Introduction

Greenbeards were originally proposed as hypothetical selfish genetic elements to illustrate how cooperation might be maintained despite the apparent cost of altruism1,2. Greenbeards have been considered unlikely because of their complexity, but empirical studies have shown the existence of various putative greenbeard types in the real world3. Nevertheless, many of these examples fall short of fulfilling all the greenbeard properties and discrepancies between theory and experiments have raised the need for additional empirical evidence4.

Two putative greenbeard examples have been described in D. discoideum. These soil amoebae propagate as unicellular organisms when food is abundant. Upon starvation, propagation stops, and the cells aggregate into a cooperative multicellular structure in which 80% of the cells become viable spores and 20% die while forming a cellular stalk5. This is a form of altruism because the stalk cells sacrifice themselves while helping in spore dispersal. The developmental process is risky because D. discoideum form chimeras that expose cells to exploitation by cheaters – strains that generate more spores than their fair share6,7,8. The developmental cell adhesion gene csaA was described as a greenbeard because its bearers cooperate with one another, whereas the absence of the gene leads to cheating9,10. This system is not a perfect greenbeard, however, because the csaA gene is not polymorphic so it does not exhibit an unusual signal that could distinguish kin from non-kin4. The tgrB1-tgrC1 allorecognition system is also a greenbeard11. tgrB1 and tgrC1 are linked polymorphic genes that encode single-pass transmembrane proteins12. There is strong evidence for their role in allorecognition in lab experiments12,13 and in nature11, but the evidence for altruism is indirect4,11. Laboratory experiments have shown that tgrB1 and tgrC1 are among the most polymorphic genes in the D. discoideum genome12,14. Together, they are necessary and sufficient for allorecognition12,13, and they function as a ligand-receptor pair in multicellular development15. Many of the laboratory experiments have used engineered strains that only differed in their tgrB1-C1 loci, but tgrB1-C1 sequence polymorphism correlates well with the segregation of wild strains, suggesting that the allorecognition observed in the laboratory is relevant in nature as well11,12,16.

The role of tgrB1-C1 in sociality has been studied in several contexts. One study showed that allorecognition can protect cooperators against cheaters caused by mutations outside the tgrB1-C1 locus17. Another showed that it protects cells from adverse interactions during slug migration11. Others have shown that the tgrB1-C1 locus defines kinship among natural isolates of D. discoideum11,12,16. The genes also have developmental roles, because they are essential for tissue integration and spore and stalk production12,15,18,19,20. Nevertheless, there has been no evidence for their direct involvement in altruism or cheating to our knowledge.

A matching pair of tgrB1 and tgrC1 encodes proteins that bind each other and mediate development and allorecognition11,15. Strains that carry two sets of different tgrB1-C1 allotypes develop well and cooperate with cells of both allotypes, suggesting that allorecognition is inclusive rather than exclusive13. In contrast, a mismatching pair of tgrB1 and tgrC1 is incompatible with normal development, and strains that carry such pairs behave like the respective null strains. A screen for genetic suppressors of such a mismatch revealed dominant mutations that activate the tgrB1 gene product21. The mutations suppressed the original tgrB1-C1 mismatch as well as mutations that inactivate tgrC1 (tgrC1–) or both genes (tgrB1–tgrC1–)15,21. Therefore, the mutant TgrB1 protein can exert its receptor activity in the absence of its ligand. The tgrB1 gene is highly polymorphic in natural populations and the proteins it encodes vary up to 13% in their amino acid sequences11,12. The polymorphism is not distributed evenly throughout the gene. The region that encodes the intracellular domain is nearly invariable, and the regions that encode the immunoglobulin folds of the extracellular domain are much less variable than the other extracellular regions12. None of the activating mutations found in the screen matched any naturally occurring SNPs in the tgrB1 coding sequences. Most of them were in the extracellular domain, and one was in the highly conserved intracellular domain. We used two of these mutations to test the hypothesis that tgrB1 activation might cause altruism, the L846F mutation that modified the intracellular domain next to S845, which is a phosphorylation site of unknown function15, and the G275D mutation that resides in the 5’ end of the region that encodes the first immunoglobulin fold12.

Here, we show that activation of the TgrB1 receptor confers altruism in that cells that carry the activated tgrB1 allele produce more of the prestalk and stalk cells when mixed with wild-type cells. The wild-type cells also produce more spores in the chimeric structures than they do in pure populations. We also show that inactivation of the tgrB1 gene causes cheating and that cheating is allotype-specific. These findings support the conclusion that tgrB1 and tgrC1 encode a green beard system that fulfills all the predicted criteria.

Results

Activation of tgrB1 confers altruism

We co-developed an RFP-marked (Red Fluorescent Protein) strain that carries the activated allele tgrB1L846F with a wild-type counterpart, GFP-marked (Green Fluorescent Protein) AX4, which has an intact pair of tgrB1-C1 alleles. The activated tgrB1 strain became enriched in the anterior (tip) and posterior (rearguard) regions at the finger stage (Fig. 1a) and in the tip during culmination (Fig. 1b). These regions normally consist of prestalk cells22,23. To further explore the relationship between the activated tgrB1 strain and its AX4 mixing partner, we expressed the activate allele tgrB1L846F in cells constitutively labeled with [act15]:CFP (Cyan Fluorescent Protein) and mixed them with AX4 cells tagged with the prestalk reporter [ecmA]:YFP (Yellow Fluorescent Protein) and the prespore reporter [cotB]:RFP. In this mix, the tgrB1L846F cells were also enriched in the prestalk regions and intermixed with the YFP-tagged AX4 prestalk cells, whereas the central prespore region was mainly occupied by the RFP-tagged AX4 prespore cells (Fig. 1c). A control mix of dual-tagged AX4 cells with constitutive CFP-labeled AX4 cells showed that the CFP-positive cells were represented throughout the developmental structure, overlapping with both the prespore and prestalk markers of the mixing partner (Fig. 1d). There were no overt differences in the prespore-prestalk ratios and positions of the labeled AX4 cells in the two mixes (Fig. 1c, d). These results suggest that the activated tgrB1 strain preferentially assumed the prestalk fate when mixed with AX4.

Fig. 1: Activated tgrB1 confers altruistic behavior.
figure 1

We used two strains that express constitutive fluorescent markers, the wild-type AX4-GFP (green) and the activated tgrB1 mutant AX4 tgrB1L846F-RFP (red). We grew the cells separately, mixed equal proportions, and co-developed them. We imaged the structures at the finger (a) and culminant stage (b) with DIC and with green and red fluorescence, and generated a merged image of the red and green channels as indicated. The brackets in panel a show the anterior region (A) that contains mainly prestalk cells and the posterior region (P) that contains mainly prespore cells. c We also co-developed constitutively labeled mutant tgrB1L846F-CFP cells (cyan) with wild-type AX4 cells carrying the prestalk reporter [ecmA]:YFP (yellow) and the prespore reporter [cotB]:RFP (red). We imaged the structures with DIC and with cyan, yellow, and red fluorescence, and generated merged images of the yellow and red (Y/R) as well as all three channels (C/Y/R) as indicated. d As a control, we co-developed constitutively labeled wild-type AX4-CFP cells (cyan) with wild-type AX4 cells carrying the same prestalk and prespore reporters and imaged them as above. e We grew wild-type AX4-GFP and mutant AX4 tgrB1L846F-RFP cells separately, developed 7×106 cells either in pure populations or mixed at equal proportions as indicated, and counted spores. The spore counts are shown as four independent replicates (symbols) and their averages (horizontal lines). The pure population counts were multiplied by 0.5 to scale them with the mixed population. Brackets and p-values (T-test, one-sided, n = 4) compare the spore counts of each strain in the two conditions. Camera settings are included in Supplementary Data 1. Source data are provided as a Source Data file.

Full size image

To test that possibility further, we developed pure and mixed strains, differentially labeled with GFP and RFP, and counted the spore production of each strain. Figure 1e shows that AX4 increased its spore production in the mix about 24% compared to its spore production in a pure population. This finding suggests that the activated tgrB1 strain altruistically increased the spore production of its mixing partner. We repeated the spore production test with another activated allele, tgrB1G275D, and found again that AX4 produced about 33% more spores in the mix than it did in the pure population (Supplementary Fig. 1). In both cases, we did not observe significant changes in the sporulation of the activated tgrB1 strain. The experiments shown in Fig. 1 suggest that tgrB1 activation confers altruistic behavior, which is manifested as increased contribution of the activated tgrB1 strain to the prestalk region and increased sporulation of the wild-type counterpart in mixed populations. These observations support the hypothesis that tgrB1 is a greenbeard element whose activation is sufficient to confer altruistic behavior.

Inactivation of tgrB1 confers cheating

If activation of tgrB1 confers altruism, its inactivation might cause cheating. To test that possibility, we co-developed differentially-labeled AX4 and tgrB1– cells, both of which carry the same tgrC1 allele. The tgrB1– cells became enriched in the central and posterior regions at the finger stage (Fig. 2a) and in the spore-bearing sorus during culmination, as well as in the basal disk (Fig. 2b). The tgrB1– cells were largely excluded from the anterior finger region (Fig. 2a) and from the culminant tip and stalk (Fig. 2b). The wild-type counterpart was the main occupant of the finger anterior (Fig. 2a), as well as the tip and stalk during culmination (Fig. 2b). These results suggest that the tgrB1– strain preferentially assumed the prespore fate whereas AX4 preferentially assumed the prestalk fate in mixed development. We then tested the effect of the tgrB1– strain on the development of the AX4 victim by mixing tgrB1– cells constitutively labeled with [act15]:CFP with AX4 cells tagged with the prestalk reporter [ecmA]:YFP and the prespore reporter [cotB]:RFP. In this mix, the tgrB1– cells were found mainly in the posterior region, intermixed with the RFP-tagged AX4 prespore cells, whereas the anterior prestalk region was mainly occupied by the YFP-tagged AX4 prestalk cells (Fig. 2c). Despite the altered distribution of the two strains, there were no obvious differences in the prespore-prestalk ratios and positions of the tagged AX4 cells in the mix. The [cotB]:RFP prespore cells occupied the posterior finger region and the [ecmA]:YFP prestalk cells occupied the anterior finger region (Fig. 2c), similar to their positions in the control mix (Fig. 1d).

Fig. 2: Inactivation of tgrB1 confers cheating behavior.
figure 2

We used two strains that express constitutive fluorescent markers, the wild-type AX4-RFP (red) and the inactivated tgrB1 mutant AX4 tgrB1–-GFP (green). We grew the cells separately, mixed equal proportions, and co-developed them. We imaged the structures at the finger stage (a) and fruiting body stage (b) with DIC and with green and red fluorescence, and generated a merged image of the red and green channels, as indicated. The green staining at the edge of the anterior region is due to reflection and is not associated with cells. c We also mixed constitutively labeled mutant tgrB1–-CFP cells (cyan) with wild-type AX4 cells carrying the prestalk reporter [ecmA]:YFP (yellow) and the prespore reporter [cotB]:RFP (red). We imaged the structures with DIC and with cyan, yellow, and red fluorescence, and generated merged images of the yellow and red (Y/R) as well as all three channels (C/Y/R), as indicated. d We grew the cells separately, developed 7 × 106 cells either in pure populations or mixed at equal proportions as indicated, and counted spores. The spore counts are shown as six independent replicates (symbols) and their averages (horizontal lines). In this case, we used 4 different alleles of AX4 tgrB1–-GFP mixed with AX4-RFP and one of AX4 tgrB1–-RFP mixed with AX4-GFP in 6 experiments. The pure population counts were multiplied by 0.5 to scale them with the mixed population. Brackets and p-values (T-test, one sided, n = 6) compare the spore counts of each strain in the two conditions. Camera settings are included in Supplementary Data 1. Source data are provided as a Source Data file.

Full size image

To further test the consequences of the interaction between the wild-type and mutant cells, we developed pure and mixed strains and counted their spore production. Figure 2d shows that the tgrB1– strain formed fewer than 2x105 spores (representing less than 3% sporulation efficiency) when developed in a pure population, but its sporulation increased more than sevenfold, to 1.4 × 106, when mixed with AX4. In addition, the presence of the tgrB1– strain caused a 24% reduction in the AX4 sporulation efficiency. We also tested the consequences of mixing the tgrB1– strain with the activated tgrB1L846F strain (Supplementary Fig. 2). In this mixing experiment, the tgrB1– strain formed 1.6 × 105 spores when developed in a pure population, but its sporulation efficiency increased more than eightfold, to 1.4 × 106, when mixed with the tgrB1L846F strain. While this effect was very similar to the mixing with AX4, the effect on the activated tgrB1L846F strain was greater. The activated tgrB1L846F spore production was reduced more than 1.5 fold, from 4.9 × 106 spores in the pure population to 3.1 × 106 spores in the mix with the tgrB1– strain (Supplementary Fig. 2).

These experiments suggest that the absence of tgrB1 results in a partial cheating behavior, which is manifested as increased propensity of the tgrB1– cells to occupy the prespore region and to produce spores. The wild-type counterpart incurred a cost that was seen as a disproportional propensity to contribute to the stalk as well as decreased spore production. This cost was even more pronounced when the counterpart expressed the activated tgrB1 allele. This partial cheating behavior is consistent with the proposed role of tgrB1 as a greenbeard element, which is necessary for altruism.

Developmental consequences of cheating by tgrB1 –

To further explore the developmental relationship between the tgrB1– cheater and the AX4 victim, we mutated tgrB1 in a strain that carries the prestalk reporter [ecmA]:GFP and the prespore reporter [cotB]:RFP. Pure tgrB1– cells do not express ecmA during development and they express very low amounts of cotB at late developmental stages but not at 16 hours, which is the wild-type finger stage (Supplementary Fig. 3). Mixing the double-tagged tgrB1– strain with unlabeled AX4 resulted in expression of both prespore and prestalk markers in the mutant at the finger stage. The tgrB1– [ecmA]:GFP prestalk cells were largely excluded from the anterior prestalk region. Instead, they were found mainly in the area that normally spans the prestalk-prespore border (Fig. 3a). The tgrB1– [cotB]:RFP prespore cells exhibited punctate staining, suggesting that only a few tgrB1– cells expressed detectable levels of the prespore marker (Fig. 3a). These cells were mainly localized in the posterior part of the finger, which is the prespore region. We also observed a narrow region of overlapping staining between the prespore and prestalk marked cells (Fig. 3a), which is not normally found in the wild type (Fig. 3b). This overlapping staining is not likely due to transdifferentiation. There is a very small number of co-labeled [cotB]:RFP and [ecmA]:GFP cells in the double-tagged wild type, and we did not observe a change in that number in the double-tagged tgrB1– strain.

Fig. 3: Cell-type development of the tgrB1– cheater and its victim.
figure 3

We grew strains separately, mixed them at equal proportions, co-developed them, and imaged at the finger stage. a We mixed unlabeled wild-type AX4 cells with mutant tgrB1– cells carrying the prestalk reporter [ecmA]:GFP and the prespore reporter [cotB]:RFP. b As a control, we mixed equal proportions of unlabeled AX4 cells with AX4 cells carrying the prestalk reporter [ecmA]:GFP and the prespore reporter [cotB]:RFP. We imaged the structures with DIC and with green and red fluorescence, and generated a merged image of the red and green channels, as indicated. Camera settings are included in Supplementary Data 1.

Full size image

As a control, we compared the expression of the reporters in the wild type and the mutant in pure populations. In the AX4 wild type, the green prestalk cells occupy the anterior region and the red prespore cells occupy the posterior region of the finger structure (Supplementary Fig. 4a). Most of the tgrB1– cells are found in loose aggregates at the same time (16 hours of development) and most of them do not express detectable levels of [cotB]:RFP and [ecmA]:GFP, with the exception of a few cells (arrows, Supplementary Fig. 4b). In very rare cases and in later times (20-24 hours of development), the tgrB1– cells form tipped aggregates and fingers that exhibit significant [ecmA]:GFP expression in the anterior and weak [cotB]:RFP expression in the posterior region (Supplementary Fig. 4c).

The results shown in Fig. 3 suggest that the cheating behavior of tgrB1– is correlated with its lack of contribution to the altruistic prestalk tissue, even though it is capable of differentiating into prestalk cells. They also suggest that tgrB1– cells receive signals from the wild type that induce them to express prespore and prestalk markers, albeit at levels lower than the wild type. Despite the altered spatial distribution of the mutant in the mixed structures, we found no evidence for enhanced transdifferentiation among the developing tgrB1– cells.

Cheating by tgrB1 – is allotype-specific

The greenbeard effect predicts that social behaviors should be allotype specific. We tested that prediction using strains with different allotypes. AX4 B1QS31C1QS31 is an AX4 strain in which the resident tgrB1-tgrC1 locus was deleted (AX4 B1ΔC1Δ) and then replaced with a tgrB1-tgrC1 locus from the incompatible strain QS3113. We also used AX4 B1AX4C1AX4, in which the replacement locus was from AX4. This strain underwent the same genetic manipulations as AX4 B1QS31C1QS31, so the two strains differ only in their allotypes13. As mixing partners, we used the respective tgrB1-deletion strains. AX4 B1ΔC1AX4 is a tgrB1-null strain in which tgrC1 is of the AX4 allotype, and AX4 B1ΔC1QS31 is a tgrB1-null strain in which tgrC1 is of the QS31 allotype. We co-developed each of the tgrB1-null strains with each of the two double-gene replacement strains and compared the sporulation efficiencies to the respective pure populations. Figure 4a shows that the AX4-type tgrB1-null strain AX4 B1ΔC1AX4 partially cheated on the compatible wild-type strain AX4 B1AX4C1AX4. This finding is similar to the one shown in Fig. 2, suggesting that the gene replacement did not alter the social interactions between the strains. Figure 4b shows that the QS31-type tgrB1-null strain AX4 B1ΔC1QS31 did not cheat on the incompatible wild-type strain AX4 B1AX4C1AX4, suggesting that an incompatible tgrC1 allele is inconsistent with cheating. Figure 4c shows that the AX4-type tgrB1-null strain AX4 B1ΔC1AX4 did not cheat on the incompatible wild-type strain AX4 B1QS31C1QS31, further supporting the hypothesis that an incompatible tgrC1 allele is inconsistent with cheating. Figure 4d shows that the QS31-type tgrB1-null strain AX4 B1ΔC1QS31 partially cheated on the compatible wild-type strain AX4 B1QS31C1QS31, further supporting the allotype-specificity hypothesis and suggesting that cheating by tgrB1-null cells is consistent with the QS31 allotype and not peculiar to the AX4 allotype. The control in Fig. 4e shows that the double-null strain AX4 B1ΔC1Δ did not cheat on the wild-type strain AX4 B1AX4C1AX4, suggesting that the absence of tgrC1 is similar to the presence of an incompatible tgrC1 in terms of cheating specificity. The experiments shown in Fig. 4 suggest that a compatible tgrC1 allele is required for social interactions in which tgrC1 encodes the allorecognition signal that directs the social behavior toward compatible kin, and tgrB1 encodes the allorecognition perception. This finding is also consistent with the developmental roles of tgrB1 as a receptor and tgrC1 as its ligand15.

Fig. 4: tgrB1– cheating is allotype-specific.
figure 4

We used strains that express constitutive GFP or RFP markers in which we deleted the resident tgrB1-tgrC1 locus (B1ΔC1Δ) and replaced it with a control locus from AX4 (B1AX4C1AX4) or a different allotype locus (B1QS31C1QS31). We also used the respective tgrB1-deletion strains (B1ΔC1AX4 and B1ΔC1QS31). In each experiment we grew the strains separately, developed 7×106 cells either in pure populations or mixed at equal proportions as indicated, and counted spores. The spore counts are shown as three or four independent replicates (symbols) and their averages (horizontal lines). The pure population counts were multiplied by 0.5 to scale them with the mixed population. Brackets and p-values (T-test, one sided, n = 3 in e and n = 4 in all the other panels) compare the spore counts of each strain in the two conditions. a Matching tgrC1 alleles from AX4 in a mix of wild-type B1AX4C1AX4 and tgrB1– mutant B1ΔC1AX4. b Non-matching tgrC1 alleles in a mix of wild-type B1AX4C1AX4 and tgrB1– mutant B1ΔC1QS31. c Non-matching tgrC1 alleles in a mix of wild-type B1QS31C1QS31 and tgrB1– mutant B1ΔC1AX4. d Matching tgrC1QS31 alleles in a mix of wild-type B1QS31C1QS31 and tgrB1– mutant B1ΔC1QS31. e A control mix of wild-type B1AX4C1AX4 and the tgrB1–tgrC1– double mutant B1ΔC1Δ. The data for pure B1AX4C1AX4 are identical between panels a and b and the data for pure and B1QS31C1QS31 are identical between panels c and d. Source data are provided as a Source Data file.

Full size image

Discussion

Previous studies have shown that the two linked genes, tgrB1 and tgrC1, encode polymorphic membrane proteins that mediate allorecognition12,13, and developmental studies showed that TgrC1 is a ligand and TgrB1 is its receptor15,19. The data shown here suggest that the tgrB1-tgrC1 locus fulfils all the criteria of a greenbeard system3,4. Our results expand on the previous findings by showing that tgrC1 encodes the perceptible greenbeard signal and tgrB1 encodes a receptor that confers altruism toward kin upon signal recognition. The previous studies provided strong support for the greenbeard hypothesis, but they linked the tgrB1-tgrC1 locus to somewhat general aspects of cooperation11,17. The findings that tgrB1 activation causes altruism and tgrB1 inactivation causes cheating against kin provide the missing direct evidence.

The altruistic action caused by activated tgrB1 was observed in two ways. First, the sporulation efficiency of the wild-type cells increased in the mixed population. Although the increase might seem modest at the 1:1 mixing ratio, it is in fact quite significant because in subsequent generations the wild-type spore proportion is expected to increase exponentially as its frequency in the population increases. Second, the activated tgrB1 cells were enriched in the prestalk region. It is somewhat surprising that we did not observe significantly reduced sporulation, but there are a few possible explanations, including differentiation of ‘differentiation-null-cells’ that would otherwise not contribute to the spores or stalks24, recruitment of ‘loners’ that would otherwise not aggregate25,26, or social effects that do not alter cell-type proportioning27. These observations were made with two different activated tgrB1 alleles, suggesting that they reflect a natural property of tgrB1 rather than an atypical new function.

The exploitation caused by tgrB1 inactivation was also manifested in two ways. First, the tgrB1– cells did not contribute significantly to the prestalk and stalk tissues in mixes with the wild type, while the wild type became the major contributor to these altruistic tissues. Second, the tgrB1– cells partially cheated by making more spores in the mix than they did in the pure population and by reducing the spore production of their wild-type partners. These behaviors satisfy the general definition of cheating, namely, benefiting from a social trait without paying the full cost28, but they differ from the prevailing definition in D. discoideum. Cheating has been defined as the production of more than the fair share of spores, where fair share was defined as the mixing ratio between the participating strains7. This definition has been expanded and refined, but it was applied mostly to facultative cheaters, which cheat in the presence of a victim but form many spores during clonal development where no victim is present29,30. The tgrB1– cells produce few spores in pure populations and they do not produce more than 50% of the spores when mixed at equal proportions with the wild type. We therefore consider them to be obligatory partial cheaters, like the fbxA– obligatory cheater that cannot sporulate in a pure population but sporulates well and cheats in the presence of a victim6. Indeed, previous studies have already shown that tgrB1– cells occupy the prespore region in mixes with the wild type, but they were not considered cheaters because of the stricter definition12. We also note that tgrC1– cells produce very few spores when mixed with the wild type18,20. This observation is consistent with the distinct but complementary roles of tgrB1 and tgrC1. As the greenbeard signal, tgrC1 is only expected to contribute to the allorecognition aspects of the system but not to affect altruism or cheating directly.

tgrB1 and tgrC1 are both essential for development and cell-type differentiation, but they have different roles and the respective null-mutants have distinct phenotypes12,15. One of the relevant differences is the null-mutants’ behavior when they are mixed with the wild type. The tgrB1– strain can cheat on the wild type because it is included in the mixed aggregates, whereas the tgrC1– mutant segregates from the wild type before tight aggregates are formed during development12. The inclusion of tgrB1– cells in the chimeric aggregate probably also explains how it can express the cotB and ecmA markers even though the greenbeard receptor function is lost. While tgrB1 and tgrC1 are essential for cell-type differentiation, they are not sufficient, and their signals can be bypassed by suppressor mutations21,31,32. In addition, D. discoideum cells employ numerous signals, both soluble and membrane-bound, to facilitate differentiation and morphogenesis33. We propose that tgrB1– cells, which are included in the chimeric aggregate, benefit from the signals produced by the wild type and can therefore produce spores and cheat. Our results therefore suggest that the greenbeard receptor function encoded by tgrB1 is required for development in a non-cell-autonomous way.

Although tgrB1 is highly polymorphic in natural populations12, the mutations described here have not been identified in the sequenced natural strains11,14. Mutations that increase altruism are likely to be eliminated from the population during evolution because they would increase the fitness of their counterparts in mixed populations. Mutations that inactivate tgrB1 cause cheating in mixed populations, but they probably get eliminated during evolution whenever the mutant cells develop clonally, due to the low sporulation efficiency of the mutant. We propose that the wild-type tgrB1 alleles confer conditional altruism, which depends on reciprocal interactions between cells with matching tgrB1-tgrC1 allotypes. This is, indeed, the property described as a greenbeard1,4.

The availability of isogenic strains that differ only in their tgrB1-tgrC1 allotype provided an opportunity to test the specificity of the greenbeard effect. The finding that the tgrC1 signal identity determined the relationship between the cheater and its mixing partner supports the greenbeard hypothesis by showing that social interactions are restricted by the allotype. The reciprocal testing with the AX4 and QS31 allotypes showed that the effects are not specific to one allotype. Moreover, the absence of cheating by the strain that lacks both tgrB1 and tgrC1 showed that a compatible signal is required for cheating. These findings also show that this greenbeard system is based on the inclusion of kin rather than the exclusion of non-kin, reaffirming previous conclusions on tgrB1-tgrC1 allotype recognition13,17. Separating the functions of the tgrC1 signal from the tgrB1 receptor revealed the role of each gene in the greenbeard locus. This separation was not possible in studies of csaA, the other D. discoideum greenbeard gene10. It also illustrates that a complex greenbeard locus can and does exist, despite early criticism that greenbeards were contrived and too complex to exist. Therefore, this study contributes significant empirical evidence to the growing body of support for the greenbeard hypothesis4, including the seminal discovery of the gp-9 greenbeard locus in fire ants34.

Quantitative and mechanistic descriptions of greenbeard systems have been made in various biological systems, but many showed only partial evidence of the required properties – a rare perceptible signal, a specific receptor, and an altruistic action toward other organisms that exhibit the same signal3,4. The tgrB1-tgrC1 system provides definitive empirical evidence for the existence of all the greenbeard properties in one locus.

Methods

Vectors

Cell-type specific markers: to generate the prestalk marker vector pDGB_A1N_ecmAp:mNeonGreen:mhcAt, we assembled the ecmA promoter, mNeonGreen coding sequence, and mhcA terminator as a transcriptional unit into the pDGB_A1N backbone using the GoldenBraid cloning method35. To generate a prespore marker vector we first cloned the cotB promoter by PCR amplification of a 1,718 bp fragment directly upstream of the cotB coding sequence from AX4 genomic DNA using the following primers: Forward: 5’ gcgccgtctc actcgggagA CATTGTGTTA TTATTTGTGT GAAAAA 3’ and Reverse: 5’ gcgccgtctc actcgcattT TTATTACTGG TACTTTTACT ATATTAATGG TATATGTATA TGAGAT 3’. The first 19 bases of the 5’ ends of each primer contain GoldenBraid-specific sequence grammar (lowercase), while the remaining bases are specific to the endogenous promoter sequence. This cotB promoter amplicon was domesticated into the pUPD2 backbone as described35. The vector pDGB_A1N_cotBp:mCherry:act8t was generated by assembling the cotB promoter, mCherry coding sequence, and actin8 terminator as a transcriptional unit into the pDGB_A1N backbone using GoldenBraid35. We also generated a single hygromycin-based vector that carries a prestalk YFP marker and a prespore RFP marker. First, we generated another prestalk marker vector by assembling the ecmA promoter, eYFP coding sequence, and mhcA terminator into the pDGB_A2N backbone. Then, we assembled it along with the pDGB_A1N_cotBp:mCherry:act8t vector into the pDGB_O1H backbone to generate pDGB_O1H_cotBp:mCherry:act8t; ecmAp:eYFP:mhcAt using GoldenBraid35.

To generate a CRISPR/Cas9 vector for mutating the tgrB1 gene, we used CRISPOR36 to design an sgRNA targeting exon 1 of tgrB1 immediately upstream of a PAM sequence found at base position 750. We annealed the following oligonucleotides and cloned into the pDGB_OH_CRISPR1 vector as described35. Sense: 5’ agaagacgga gcaCCAAAGC TCGATAAAAT GGA gtttcc gtcttct 3’; antisense: 5’ agaagacgga aacTCCATTT TATCGAGCTT TGGtgctccg tcttct 3’. The 13 bases at either end of the oligonucleotide (lowercase) contain GoldenBraid-specific grammar, while the intervening 20 bases (uppercase) constitute the guide sequence.

We also used the published vectors pDXA-tdTomato13, ptgrB1:TgrB1AX4(L846F) bsR15, ptgrC1:TgrC1AX4 bsR15 and ptgrC1:TgrC1QS31 bsR15. Other fluorescent protein expression vectors, pDXA-mCherry, pDM304-mCherry, and pDXA-mCerulean, in which the fluorescent marker gene was driven by the actin15 promoter, were a kind gift from Shigenori Hirose.

Strains and strain construction

All the D. discoideum strains were generated by transformation of AX437 or its derivatives as detailed in Supplementary Table 1.

Cell growth and transformation

We maintained D. discoideum cells at 22 °C in HL5 medium in a submerged culture and grew them for transformation and development in shaking suspension at 200 RPM with the adequate supplements and antibiotics12. We transformed the cells by electroporation, cloned by plating in association with bacteria and identified the desired clones by PCR analysis13. Before each experiment, we grew the cells at the logarithmic phase without antibiotics for 24 hours. Mutagenesis by CRISPR/Cas9 was performed as described35.

We validated all the transformed strains by PCR and sequencing of the relevant genes. The PCR primers and sequencing oligonucleotides used for each diagnostic amplification, and sequencing oligonucleotides for CRISPR validation are listed in Supplementary Table 2.

Development, imaging, and mixing experiments analysis

We induced development by washing the cells twice in KK2 (20 mM potassium phosphate, pH6.4) followed by starvation in a humid chamber at 22 °C. In mixing experiments, we grew the strains separately, washed the cells separately, counted them, and mixed in equal proportions before depositing them on solid substrates for development. To image developmental structures, we plated cells at a density of 2–5x105 cells/cm2 on 1.5% Nobel agar made in KK2. Fluorescence and Differential Interference Contrast (DIC) microscopy images were captured with a Nikon (Tokyo, Japan) Eclipse Ti microscope using the NIS Elements 4.51.00 software35. The images shown are representative of at least 2 independent replications, each consisting of several hundred structures. To measure sporulation and cheating, we developed the cells on black nitrocellulose filters for 40 hours and harvested spores as described38 with the following exceptions: we deposited 1.4x106 cells/cm2 cells on each quarter filters and placed three replicate quarter filters on one filter pad per sample. After spore collection, we counted the spores and captured fluorescence and DIC micrographs of several fields to calculate the sporulation efficiency of each strain in the mix. The average of the three-quarter filter counts was reported as one data point. Each experiment was repeated at least three independent times. To compare between development in pure population and development in 1:1 mixes, we multiplied the pure population spore counts by 0.5 to scale them with the mixed populations. We performed one-sided paired T-tests using Microsoft Excel version 16.82 to compare pairs of pure and mixed populations.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.