Altruism and social cheating in the social amoeba ... - Nature

letters to nature

................................................................. Altruism and social cheating in the social amoeba Dictyostelium discoideum Joan E. Strassmann, Yong Zhu & David C. Queller Department of Ecology and Evolutionary Biology, Rice University, PO Box 1892, Houston, Texas 77251-1892, USA ..............................................................................................................................................

The social amoeba, Dictyostelium discoideum, is widely used as a simple model organism for multicellular development1,2, but its multicellular fruiting stage is really a society. Most of the time, D. discoideum lives as haploid, free-living, amoeboid cells that divide asexually. When starved, 104 ±105 of these cells aggregate into a slug. The anterior 20% of the slug altruistically differentiates into a non-viable stalk, supporting the remaining cells, most of which become viable spores3±5. If aggregating cells come from multiple clones, there should be selection for clones to exploit other clones by contributing less than their proportional share to the sterile stalk. Here we use microsatellite markers to show that different clones collected from a ®eld population readily mix to form chimaeras. Half of the chimaeric mixtures show a clear cheater and victim. Thus, unlike the clonal and highly cooperative development of most multicellular organisms, the development of D. discoideum is partly competitive, with con¯icts of interests among cells. These con¯icts complicate the use of D. discoideum as a model for some aspects of development, but they make it highly attractive as a model system for social evolution. The problem of potential con¯ict among multiple clones within fruiting bodies has been recognized6±13, but little investigated. The altruistic stalk cells give up reproduction in order to bene®t the spore cells, by lifting them above the hazards of the soil or increasing their chances of dispersal to a more favourable environment7,14. When fruiting bodies contain only one clone, like multicellular animals, sterile stalk cells are favoured by kin selection15; genes for facultative sterility can spread if they help copies of themselves in the reproductive cells. However, if D. discoideum aggregates include more than one clone, they may be more analogous to the societies of social insects, which have sterile castes, but also have multiple genotypes and genetic con¯icts of interests7. Selection within individual fruiting bodies could favour cheater clones that contribute less than their fair share of the sterile stalk6±13. Such con¯icts would be averted if Dictyostelid clones always multiply and aggregate in isolation from other clones6,8 but, in species where it has been examined, genetically diverse clones do sometimes occur in close proximity in nature16,17. Within-individual selection and developmental con¯ict could be minimized in two other ways. First, cells might recognize and exclude foreign clones9, as occurs in many marine invertebrates18. If this does not occur, there might still be some mechanism that assures fairness in chimaeric mixtures9 analogous to the way meiosis normally ensures fairness in segregation. There has been no systematic study of these points in Dictyostelids, particularly in natural populations. Chimaeras are sometimes generated, usually with laboratory strains19±21. However, recognition abilities could have been lost in a laboratory environment where they are not needed. Moreover, because many of these laboratory mixtures involve a developmental mutant and its otherwise isogenic parent clone, they would not reveal a genetic recognition system based on genetic differences at other loci. Several such chimaeras show unequal apportionment of the two clones to spore and stalk19±21, but it is not known if such cheating is common in nature. Buss9 found a natural stalkless clone of Dictyostelium mucoroides that could exploit the stalk-forming ability of a neighbouring clone. It did NATURE | VOL 408 | 21/28 DECEMBER 2000 | www.nature.com

not mix with other clones in the vicinity, suggesting that the parasite might have been a new mutant that could only victimize its parent. To test whether clones generally recognize and exclude each other, we obtained clonal isolates of D. discoideum collected from a natural population at the type locality in North Carolina. We performed pairwise mixing experiments, testing for chimaeric slugs by amplifying microsatellite loci whose alleles differed in the two clones. The presence of both parental alleles in a slug indicates chimaera formation (Fig. 1). All 15 pairs of clones tested consistently formed chimaeras. This failure to exclude non-relatives renders D. discoideum clones potentially vulnerable to cheating. A second experiment shows that chimaeric mixtures are often not fair, in the sense that the clonal composition of spores differs from that of stalks. We made 12 mixtures of two clones each. Instead of genotyping entire slugs, we sampled the prespore and prestalk regions of each slug, using the fact that the anterior 20% of migrating slugs will form the stalk. For each sample, we ampli®ed a variable microsatellite locus, ran it out on a gel to separate the two distinct parental alleles, and quanti®ed the amount of the two alleles. Because the two alleles are ampli®ed in the same reaction and have the same primer binding sites, the polymerase chain reaction (PCR) products should be proportional to the relative numbers of target sequences, that is, to the relative cell numbers of the two clones. These relative counts are suf®cient to test the null hypothesis of fair behaviour, which is that a clone's representation in the stalk should be the same as its representation in the spores. For example, if a clone contributes 70% of the spores, it should also contribute 70% of the stalk. We note that we do need both numbers; 70% representation in spores is not by itself evidence for cheating because it could be due to other factors like differential growth rates before fruiting. The null hypothesis is rejected in 6 of the 12 mixtures (Fig. 2), indicating that cheating by one clone is common. The results so far are consistent with two forms of cheating. They appear to suggest that one clone often gains by producing more spore cells than it normally would, exploiting the other clone's stalk cells. However, unequal representation in the stalk and spores could also arise from different solitary allocations, unchanged in mixture. For example, if clone A always has a stalk/spore allocation of 30/70 and clone B always has 10/90, in a 50:50 mixture clone A will contribute only 1/4 (that is, 10/(10+30)) of the stalk cells, while obtaining 9/16 (that is, 90/(70+90)) of the spores. In this ®xed allocation model, the cheater gets no more spores in mixture than it does alone, but it may gain another advantage. It will have its spores at a greater height in mixture than alone. Similarly, its victim clone will have its spores on a shorter stalk in mixture than alone. The cheater's cost-free increase in stalk height is likely to be advantageous, provided stalk heights have been previously selected to re¯ect

s c1 c2 m1 m2 m3 m4 m5

NC28.1

s c1 c2 m1 m2 m3 m4 m5

NC41.2 NC67.2

NC28.2

Figure 1 Microsatellite genotypes of slugs from two-clone mixture experiments. For each experiment, lanes c1 and c2 are slugs from the two clones cultured alone (labelled at the side with the clone name), each showing a single allele, as expected for haploid D. discoideum. Lanes m1±m5 show ®ve slugs derived from mixing amoebae of the two clones. Each is chimaeric, showing both alleles. Lane s is a size marker. All 15 clone pairs tested showed clear mixing, with an average of 21 slugs tested. (When contributions of the two parent clones appeared unequal, as for NC 28.1 and NC 28.2 on the left, the inequalities tended to be similar across slugs, consistent with differences in growth rates before aggregation, or with differences in size or viability of the sori used to initiate the mixture.)

© 2000 Macmillan Magazines Ltd

965

letters to nature a balance between bene®t and cost. When stalks are costly, their average length would be set below the height that would yield maximum bene®ts to spores. Cheaters gain such height bene®ts by allowing the victims to pay more of the costs. Buss's9 stalkless mutant was apparently a cheater of this form, allocating no cells to stalk when alone or in mixture. Distinguishing between these two forms of cheating requires estimates of solitary allocation to spores and stalks. Direct cell counts are not feasible for stalks, so we used two different indices of relative allocation. For the ®rst, we measured 15 randomly chosen fruiting bodies of each clone. As a measure of a clone's relative allocation to spores, we took the mean of S1, where S1 = (sorus width)/(sorus width + stalk length), where the sorus is the spherical cluster of spore cells on top of the stalk. Our second relative estimator of spore allocation, S2, was the length of a slug's prespore region divided by the length of the slug. Each clone's mean was estimated from 20 randomly chosen slugs of each clone stained with neutral red, which dyes the anterior prestalk region. For each of the mixtures in Fig. 2, we then calculated a measure of how much

Clone Clone 1 2 Prespore Prestalk NC60.1

NC60.2

NC105.1

NC34.1

NC28.1

NC28.2

NC63.2

NC69.1

NC98.1

NC70.1

NC101.1

NC99.1

NC94.2

NC94.1

NC85.1

NC85.2

NC75.2

NC28.1

NC4

NC47.2

NC34.1

NC34.2

NC85.2

NC39.1

0.0

0.2

0.4

0.6

0.8

1.0

Proportion of clone 1

Figure 2 Relative contributions of mixed clones to prespore and prestalk regions of chimaeras. Each of the 12 four-pattern bars shows the averages from seven replicates of one pairwise mixture of two clones. The legend at the top shows that the top half (speckled) represents the prespore region and the bottom half (solid) represents the prestalk region. Each of these is divided according the contributions of the clone on the left (dark) and the clone on the right (light). For example, in the mixture of clones 60.1 and 60.2, clone 60.1 gets 53% of the prespore cells (dark speckled), but contributes only 3% of the prestalk (dark solid). This disparity makes 60.1 a cheater, gaining spores without paying its share of the cost. The null hypothesisÐthat a clone contributes equal percentages to the prespore and prestalk regionsÐwould be represented by equal-sized dark portions in both top and bottom bars, as in the legend at the top. It is rejected for the top nine mixtures taken individually (t-tests, paired by slug, of arcsine-squareroot transformed proportions), and for the top six mixtures when corrected for multiple comparisons (step-down Holm adjustment)30. 966

more (or less) the sel®sh clone allocates to spores than its victim does, when each is alone: D1 = S1(sel®sh) - S1(victim), and D2 = S2(sel®sh) - S2(victim). The ®xed allocation hypothesis predicts that these measures should be positive (the more ``spory'' clone should be the sel®sh one in mixture), but D1 was positive in only 4 of 12 cases, and D2 in 6 of 12. It also predicts that D1 and D2 should be positively related to the magnitude of sel®shness in mixture (as measured by the sel®sh clone's spore proportion minus its stalk proportion in Fig. 2). The prediction fails; both regressions are strongly and signi®cantly negative (both slopes = 0.59, P , 0.05) indicating that typically higher allocation to spores alone means less sel®shness in mixture. Thus, there is no evidence that cheating is restricted to the ®xed-allocation advantage of having a higher stalk. Instead, it must often involve spore number advantages. Certainly ®xed allocation cannot explain extreme results such as the 60.1/60.2 mixture (Fig. 2); clone 60.1's allocation in mixture would imply that is has virtually no stalk when alone, but in fact it has a normal stalk. The ability to cheat appears to be common in this natural population, something that has not been demonstrated for other examples of developmental cheating in a social bacterium22, a colonial ascidian23, and Dictyostelium9,20,21. Most previous examples come from laboratory mutants20,21,22 whose signi®cance in natural populations is unknown. In examples from natural populations, the incidence of cheating appears low: ascidians usually exclude each other24, and only one natural stalkless mutant of D. mucoroides has been found9. How cheating is maintained at such high levels in D. discoideum is not yet known. Factors that may be important are the frequency of between-clone encounters in the ®eld, whether a clone's ability to cheat is general or speci®c to certain victim clones, and costs of being a cheater. The prevalence of cheating in D. discoideum complicates its use as a model system for development of organisms like vertebrates, which are clonal and essentially free from such con¯icts. Within-cell processes may be little affected by this difference, but between-cell interactions involving signalling, adhesion and differentiation may be very different. Signalling among non-relatives often involves elements of deception and manipulation25, as when female spiders performing courtship behaviours eat the male instead of mating with him26. Recognizing that between-cell signalling in D. discoideum may have similar con¯ict elements might help resolve the long-standing problem of which cells go to spore and stalk, much as recognition of worker±queen con¯ict was essential for understanding social insect sex ratios27. These con¯icts of interest might even account for the very feature that has made Dictyostelium attractive as a model developmental system: its simplicity. Social amoebae appear to possess most of the kinds of molecular mechanisms required to evolve more complex forms of multicellularity, but they have not done so28. Perhaps what they lack most is an effective means to suppress con¯ict, either by assuring genetic uniformity or by enforcing fairness in mixtures. Biologists interested in social evolution have long been hampered by the dif®culties of studying organisms like social insects. Selection experiments are rarely feasible, and there is little foundation for identifying the genes that affect sociality and tracing their effects. In contrast, such studies should be facilitated in D. discoideum by its short generation time, the imminent completion of the genome sequence29, and decades of careful work on its cell biology and its molecular biology, making it an invaluable model system for evolutionary and mechanistic studies of altruism and cheating. M

Methods

We obtained natural clonal isolates of D. discoideum collected from the type locality, Little Butts Gap, North Carolina. These were collected from a 100 m ´ 50 m wooded area (clones 26±75) or within 1,500 m of this area (clones 76±105) and had previously been shown to be genetically diverse, sometimes within the same small soil sample16. We also used the original type clone, NC4. Microsatellite sequences, which are tandem repeats of short DNA motifs, were located


NATURE | VOL 408 | 21/28 DECEMBER 2000 | www.nature.com

letters to nature in GenBank and from the Genome Sequencing Centre Jena (http://genome.imb-jena.de/ dictyostelium/). Primers were designed to amplify the repeat sequence (information is available from the authors on request), which is often variable in length. To determine if clones mix, we performed 15 pairwise mixing experiments. For each, we grew up fruiting bodies for the two clones separately, picked 5 sori from each, and mixed all 10 with Klebsiella aerogenes as a food source in 0.2 ml of sterile distilled water, where they dispersed as hundreds of thousands of independent amoebae. We spread this homogeneous mixture of two clones out on an SM/5 plate. After several days growth at 22 8C, the starving amoebae aggregated and formed migrating slugs, and 24 were chosen for genotyping at a microsatellite locus chosen because the two parent clones possessed different alleles. Mixing was con®rmed if a slug showed both parental alleles. We mixed all the clone pairs listed in Fig. 2 except NC94.1-NC94.2 (which did not grow in this trial), plus four others: NC41.2-NC67.2; NC54.1-NC54.2; NC42.1-NC43.1; and NC66.2-NC 85.1. We use the original nomenclature16. To assess the frequency of cheating in mixtures, we made 12 pairwise mixtures (listed in Fig. 2) as in the ®rst experiment. For each mixture, we selected 7 migrating slugs. The front 10% of the slug was excised as a sample of the prestalk region. The prespore region of each slug was sampled twice, using the posterior 10% of the slug and an equivalent portion from the middle of the prespore region. We extracted DNA separately from all three portions and ampli®ed a microsatellite locus that differed for the two parent clones, incorporating 35S-labelled dATP in the PCR reaction. We ran the product DNA out on a 6% polyacrylamide gel. We then quantifed the relative amounts of the two parental alleles using a phosphorimager that measures the radiation given off by any region of the gel. To remove background, we subtracted the average radiation reading of two other areas, located above and below the bands, and identical in size to the band. Exposures were kept low to avoid saturation effects. We compared the prestalk with the average of the two prespore samples. A replicate set of the same 12 mixtures was conducted, using only the posterior 10% of the slug for the prespore region, and 11 of the 12 showed differences in the same directions as in Fig. 2. Marker loci were chosen such that the two alleles did not have overlapping stutter bands, but were within a few repeats of each other to avoid differential ampli®cation. However, even if the one of the two alleles ampli®ed better, it would do so consistently in prestalk and prespore samples, so that a test for a difference between these two samples remains valid. We con®rmed that the method works by amplifying a second locus for ®ve of our mixtures. For each, the correlation across slugs of two locus-speci®c estimates of %prestalk minus %prespore was highly signi®cant (P , 0.005) with the correlation averaging 0.8. If the results we reported were due to differential ampli®cation or some other artefact, we would expect no correlation, or negative correlations as often as positive.

Acknowledgements We thank R. Gomer for advice and training; D. Welker for supplying the clones; J. Keay, W. Castle, S. Reddy and J. Damon for assistance with laboratory work; and D. Rozen, R. Kessin, G. Velicer and J. Bonner for comments on the manuscript. This work was supported in part by the US National Science Foundation. Correspondence and requests for materials should be addressed to J.E.S. (e-mail: [email protected]).

................................................................. Asymmetric leaves1 mediates leaf patterning and stem cell function in Arabidopsis

Mary E. Byrne*, Ross Barley², Mark Curtis*³, Juana Maria Arroyo*, Maitreya Dunham*³, Andrew Hudson² & Robert A. Martienssen* * Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, New York 11724, USA ² Institute of Cell and Molecular Biology, University of Edinburgh, Edinburgh EH9 3JR, UK ..............................................................................................................................................

Received 15 September; accepted 18 October 2000. 1. Maeda, Y., Inouye, K. & Takeuchi, I. (eds) DictyosteliumÐA Model System for Cell and Developmental Biology (Universal Academy, Tokyo, 1997). 2. Gross, J. D. Developmental decisions in Dictyostelium discoideum. Microbiol. Rev. 58, 330±351 (1994). 3. Bonner, J. T. The Cellular Slime Molds (Princeton Univ. Press, Princeton, 1967). 4. Raper, K. B. The Dictyostelids (Princeton Univ. Press, Princeton, 1984). 5. Raper, K. B. Pseudoplasmodium formation and organization in Dictyostelium discoideum. Journal of the Elisha Mitchell Scienti®c Society 56, 241±282 (1940). 6. Williams, G. C. Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought (Princeton Univ. Press, Princeton, 1966). 7. Gadagkar, R. & Bonner, J. T. Social insects and social amoebae. J. Biosci. 19, 219±245 (1994). 8. Armstrong, D. P. Why don't cellular slime molds cheat. J. Theor. Biol. 109, 271±283 (1984). 9. Buss, L. W. Somatic cell parasitism and the evolution of somatic tissue compatibility. Proc. Natl Acad. Sci. USA 79, 5337±5341 (1982). 10. Hilson, J. A., Kolmes, S. A. & Nellis, L. F. Fruiting body architecture, spore capsule contents, sel®shness, and heterocytosis in the cellular slime mold, Dictyostelium discoideum. Ethol. Ecol. Evol. 6, 529±535 (1994). 11. Wilson, D. S. & Sober, E. Reviving the superorganism. J. Theor. Biol. 136, 337±356 (1989). 12. Kessin, R. H. in DictyosteliumÐA Model System for Cell and Developmental Biology (eds Maeda, Y., Inouye, K. & Takeuchi, I.) 3±13 (Universal Academy, Tokyo, 1997). 13. Atzmony, D., Zahavi, A. & Nanjundiah, V. Altruistic behaviour in Dictyostelium discoideum explained on the basis of individual selection. Curr. Sci. 72, 142±145 (1997). 14. Huss, M. J. Dispersal of cellular slime moulds by two soil invertebrates. Mycologia 81, 677±682 (1989). 15. Hamilton, W. D. The genetical evolution of social behaviour. I, II. J. Theor. Biol. 7, 1±52 (1964). 16. Francis, D. & Eisenberg, R. Genetic structure of a natural population of Dictyostelium discoideum, a cellular slime mold. Mol. Ecol. 2, 385±391 (1993). 17. Ketcham, R. B. & Eisenberg, R. M. Clonal diversity in populations of Polysphondylium pallidum, a cellular slime mold. Ecology 70, 1425±1433 (1989). 18. Grosberg, R. K. The evolution of allorecognition speci®city in clonal invertebrates. Q. Rev. Biol. 63, 377±412 (1988). 19. Filosa, M. F. Heterocytosis in cellular slime molds. Am. Nat. 96, 79±91 (1962). 20. Houle, J., Balthazar, J. & West, C. M. A glycosylation mutation affects cell fate in chimeras of Dictyostelium discoideum. Proc. Natl. Acad. Sci. USA 86, 3679±3683 (1989). 21. Ennis, H. L., Dao, D. N., Pukatzki, S. U. & Kessin, R. H. Dictyostelium amoebae lacking an F-box protein form spores rather than stalk in chimeras with wild type. Proc. Natl Acad. Sci. USA 97, 3292± 3297 (2000). 22. Velicer, G. J., Kroos, L. & Lenski, R. E. Developmental cheating in the social bacterium Myxococcus xanthus. Nature 404, 598±601 (2000). 23. Stoner, D. S., Rinkevich, B. & Weissman, I. L. Heritable germ and somatic cell lineage competitions in chimeric colonial protochordates. Proc. Natl Acad. Sci. USA 96, 9148±9153 (1999). 24. Grosberg, R. K. & Quinn, J. F. The genetic control and consequences of kin recognition by the larvae of a colonial marine invertebrate. Nature 322, 456±459 (1986). 25. Dawkins, R. & Krebs, J. R. in Behavioural Ecology: An Evolutionary Approach 2nd edn (eds Krebs, J. R. & Davies, N. B.) 282±309 (Blackwell, Oxford, 1978).

NATURE | VOL 408 | 21/28 DECEMBER 2000 | www.nature.com

26. Elgar, M. A. & Nash, D. R. Sexual cannibalism in the garden spider Araneus diadematus. Anim. Behav. 36, 1511±1517 (1988). 27. Queller, D. C. & Strassmann, J. E. Kin selection and social insects. Bioscience 48, 165±175 (1998). 28. Loomis, W. F. Four Billion Years; An Essay on the Evolution of Genes and Organisms (Sinauer, Sunderland, Massachusetts, 1988). 29. Kay, R. R. & Williams, J. G. The Dictyostelium genome projectÐan invitation to species hopping. Trends Genet. 15, 294±297 (1999). 30. Holm, S. A simple sequentially rejective multiple test procedure. Scand. J. Statist. 6, 65±70 (1979).

Meristem function in plants requires both the maintenance of stem cells and the speci®cation of founder cells from which lateral organs arise. Lateral organs are patterned along proximodistal, dorsoventral and mediolateral axes1,2. Here we show that the Arabidopsis mutant asymmetric leaves1 (as1) disrupts this process. AS1 encodes a myb domain protein, closely related to PHANTASTICA in Antirrhinum and ROUGH SHEATH2 in maize, both of which negatively regulate knotted-class homeobox genes. AS1 negatively regulates the homeobox genes KNAT1 and KNAT2 and is, in turn, negatively regulated by the meristematic homeobox gene SHOOT MERISTEMLESS. This genetic pathway de®nes a mechanism for differentiating between stem cells and organ founder cells within the shoot apical meristem and demonstrates that genes expressed in organ primordia interact with meristematic genes to regulate shoot morphogenesis. The shoot apical meristem (SAM) comprises slowly dividing stem cells at the centre and daughter cells at the periphery, from which organ founder cells are recruited. Founder cells divide rapidly, initiating the outgrowth of organ primordia while polarity is established along the proximodistal, dorsoventral and mediolateral axes1,2. The mechanism by which stem cell and founder cell derivatives are distinguished is obscure, but is likely to involve a highly conserved class of homeobox genes related to KNOTTED1 in maize (KNOX genes). KNOX genes are expressed in the SAM but are downregulated in founder cells at the time of leaf initiation3. They are implicated in maintaining division or preventing differentiation of cells in the SAM. Loss-of-function mutations in the Arabidopsis KNOX gene SHOOT MERISTEMLESS (STM) result in embryos that lack a SAM4,5. Recessive mutations in the kn1 gene of maize are also defective in meristem maintenance6. In contrast, gain-of-function mutations that result in ectopic expression of KNOX genes in maize disrupt normal leaf development causing distal displacement ³ Present address: Institute of Biological Sciences, University of Wales, Aberystwyth, UK (M.C.); Genetics Department, Stanford, California 94305, USA (M.D.).


967