J Biosci Vol. 43, No. 5, December 2018, pp. 1025–1036 Ó Indian Academy of Sciences DOI: 10.1007/s12038-018-9810-1

Species recognition in social amoebae

1,2 1 IKUMI SHIBANO HAYAKAWA * and KEI INOUYE 1Department of Botany, Graduate School of Science, Kyoto University, Kyoto 606-8502, Japan 2Department of Physics, Graduate School of Science, Kyoto University, Kyoto 606-8502, Japan

*Corresponding author (Email, [email protected])

MS received 9 May 2018; accepted 12 September 2018; published online 6 October 2018

Aggregative multicellularity requires the ability of cells to recognise conspecifics. Social amoebae are among the best studied of such organisms, but the mechanism and evolutionary background of species recognition remained to be investigated. Here we show that heterologous expression of a single purpureum gene is sufficient for D. discoideum cells to efficiently make chimaeric fruiting bodies with D. purpureum cells. This gene forms a bidirectional pair with another gene on the D. purpureum genome, and they are both highly polymorphic among independent wild isolates of the same mating group that do not form chimaeric fruiting bodies with each other. These paired genes are both structurally similar to D. discoideum tgrB1/C1 pair, which is responsible for clonal discrimination within that species, suggesting that these tgr genes constitute the species recognition system that has attained a level of precision capable of discriminating between clones within a species. Analysis of the available genome sequences of social amoebae revealed that such gene pairs exist only within the clade composed of species that produce precursors of sterile stalk cells (prestalk cells), suggesting concurrent evolution of a precise allorecognition system and a new ‘worker’ cell-type dedicated to transporting and supporting the reproductive cells.

Keywords. Chimaera; Dictyostelium purpureum; evolution; interspecific recognition; social amoeba; tgr genes

1. Introduction 2011). Since the coexistence of different species of social amoebae is common in nature (Eisenberg 1976; Sathe et al. The ability of cells to discriminate self from non-self is of 2010), the importance of self-/non-self-discrimination would fundamental importance for the existence of species. Uni- be particularly high in order to avoid chimaeric fruiting body cellular organisms need to distinguish their mates from formation. others in reproduction and predation (Sakaguchi et al. 2001; Use of different chemotactic agents is one way to avoid Seike et al. 2015). In multicellular organisms, self-/non-self- mixing of different species, as is the case for distantly related recognition is also crucial to successful reproduction and species (Konijn et al. 1968; Shimomura et al. 1982; van defence, with the added complexity of the separation of Haastert et al. 1982; de Wit and Konijn 1983; Asghar et al. reproductive and non-reproductive cell lineages (Litman 2012). Cells of closely-related species sharing a common et al. 2005; Buchmann 2014). Social amoebae have a unique chemoattractant, such as Dictyostelium discoideum and D. life cycle that alternates between unicellular and multicel- purpureum both using cAMP as the chemoattractant, would lular phases, the latter being initiated by aggregation of aggregate together to form a chimaeric structure, but they single cells by chemotaxis. The final form of the multicel- largely segregate from each other to make separate fruiting lular structure in asexual development is the fruiting body, in bodies (Raper and Thom 1941; Bonner and Adams 1958; which only spores survive while stalk cells undergo cell Jack et al. 2008; Sathe et al. 2013). death. This type of multicellularity considerably increases Kin discrimination has been shown to occur even within a the risk of genetically heterogeneous cells being mixed species (Mehdiabadi et al. 2006; Ostrowski et al. 2008). The together, and any difference in their propensity to become genetically tractable model species D. discoideum has served spores directly affects their fitness, posing a potential hazard as a tool to investigate the molecular mechanisms of kin to species stability (Atzmony et al. 1997; Strassmann et al. recognition, and a pair of polymorphic recognition mole- 2000; Nanjundiah and Sathe 2011; Strassmann and Queller cules Tiger B1 (TgrB1) and Tiger C1 (TgrC1) have been

Electronic supplementary material: The online version of this article (https://doi.org/10.1007/s12038-018-9810-1) contains supplementary material, which is available to authorized users. http://www.ias.ac.in/jbiosci 1025 1026 IS Hayakawa and K Inouye identified as responsible for the segregation of different transformants derived therefrom were grown in axenic clones (Benabentos et al. 2009; Hirose et al. 2011). How- medium consisting of Proteose Peptone 2 (Difco) 14.3 g, ever, these studies have not been extended to other species, Yeast extract (Difco) 7.15 g, Na2HPO4Á12H2O 1.28 g, and nothing is known about the mechanism of between- KH2PO4 0.49 g, H2O 1000 ml (Watts and Ashworth 1970) species recognition, nor has the question been addressed of in culture dishes or with shaking at 175 rpm. The axenic how the mechanisms of within- and between-species medium was supplemented with G418 (20 lg/mL) for recognition may be related (Nydam and De Tomaso 2011). growth of transformant cells. All growth and development The purpose of this study is to identify factors involved in were conducted at 22°C unless otherwise indicated. interspecific recognition using the two species D. dis- coideum and D. purpureum as a representative model, and to gain insight into the relationship between the intra- and 2.3 Sequencing and assembly interspecific recognition systems. We will further discuss to what extent the features found in the recognition systems of Vegetative cells of P1Ba clone 2a were harvested after these species are shared in the social amoebae phylogeny. growth on SM plates in association with Klebsiella aero- genes, extensively washed to remove bacteria, and starved for *3 hours to allow digestion of remaining bacteria. 2. Material and methods Genomic DNA was prepared using GenEluteTM Genome DNA Miniprep Kit (Sigma-Aldrich, St. Louis, MO, USA), 2.1 Strains followed by RNAse treatment. The integrity of DNA was checked by agarose gel electrophoresis. Genome sequencing We mainly used D. purpureum strain P1Ba and D. dis- was conducted on the Illumina HiSeq platform by the coideum strain AX2. Within the morphological species D. paired-end method with 100-bp reading (Hokkaido System purpureum, there are distinct phylogenetic groups, which can Science, Sapporo, Japan). After removal of adapter be distinguished by mating compatibility and morphological sequences with Skewer v.0.1.123, the reads aligned to the features (Hagiwara et al. 2004) as well as DNA sequences Klebsiella aerogenes genome (Shin et al. 2012) were (Mehdiabadi et al. 2009). P1Ba belongs to the temperate removed using bwa-0.7.12, and the remaining reads were form of Hagiwara et al. (2004), as judged from mating assembled using Velvet-1.2.10 (Zerbino and Birney 2008; experiments, and to group B of Mehdiabadi et al. (2009) from Zerbino et al. 2009) with various k-mer values. The contigs rDNA sequences. Compared with the sequenced strain of D. shown in this report are from the assembly with a k-mer of purpureum, QSAX1 (Sucgang et al. 2011), P1Ba showed a 67. lower level of diversification of the tgr gene family (approx. 10 in P1Ba, 37 in QSAX1). Additional 11 wild isolates belonging to the same mating group as P1Ba were also 2.4 Data analysis examined (supplementary tables 1 and 2). Four of these isolates, including TNS-C-7 (= ‘Hagiwara 5’ in Hagiwara The assembled sequences were converted into a blast data- et al. (2004), were obtained from National BioResource base, and tblastn searches were performed for sequences Project Cellular Slime Molds (http://nenkin.nbrp.jp/), while related to all of the 37 D. discoideum genes annotated as tgr the others (except the one of unknown origin) were collected on dictyBase (tgrA1 *tgrS1) and csaA. Sequence searches from natural populations at various locations in Japan (sup- were performed using ncbi-blast-2.2.26?. The output was plementary table 1). Outside group 4, fila- mapped to the contigs using perl scripts and gnuplot, and mentosum, D. lacteum, D. rhizopodium and D. laterosorum bidirectional gene pairs of which both genes show E-values were obtained from Dicty Stock Center (http://dictybase.org/ less than 1e-10 were further investigated to identify candi- StockCenter/StockCenter.html). subglobosum date paired tgr genes. For comparative analysis, the genome 2CBg, A. amazonicum 2B1Ba, P. pallidum 2F2Ca, D. poly- sequences listed below were searched for potential tgr genes. cephalum 2A2j and 1JNd1, D. menorah 1Za32a, P. vio- Sequence alignments were performed using E-INS-i in laceum P4Ba and 1MN8b, and D. giganteum E1SS12 were MAFFT (Katoh et al. 2005; Katoh and Standley 2013), and collected at various sites in Japan. phylogenetic trees were constructed using RAxML (Sta- matakis 2014) and visualised with FigTree (http://tree.bio. ed.ac.uk/software/figtree/). Sequence polymorphism of the 2.2 Culture candidate tgr genes was examined using additional wild isolates of the same mating group as P1Ba. The respective Wild isolates of D. purpureum were grown in association tgr regions were cloned by PCR, sequenced on an with Klebsiella aerogenes or E. coli B/r and washed free of ABI3130xl Genetic Analyzer (Applied Biosystems), and the bacteria by repeated centrifugation. D. discoideum AX2 and numbers of synonymous and non-synonymous substitutions Species recognition in social amoebae 1027 were calculated by the method of Nei and Gojobori (1986) 4 pulses with 10 sec intervals, 22°C). The cell suspension using SNAP.pl (Korber 2000). Pfam domain searches were was diluted in PB and gently washed three times. Rho- performed using Kyoto Encyclopedia of Genes and Gen- damine-labelled D. purpureum cells were mixed with freshly omes (KEGG) database (http://www.genome.jp/tools/motif/). harvested D. discoideum transformant cells at a ratio of 1:1, The pairwise phylogenetic distance between species/strains and the cell suspension (107 cells/ml) was spotted on a non- was estimated by calculating the branch length on a tree nutrient agar plate (1.25%). After incubation at 10°C about based on the published phylogeny constructed 24 hours, aggregating cells were photographed using a with SSU rDNA sequences (Romeralo et al. 2011) using the confocal laser microscope. For mixing experiments of other perl script ‘distontree.pl’ written by Dr. Akifumi S. Tanabe species, cells were labelled with tetramethylrhodamine- (http://www.fifthdimension.jp/products/distontree/). For dextran or fluorescein-dextran (Sigma) by electroporation. In strains not included in the original tree, local trees based on some experiments using non-group 4 species, SSU rDNA sequences were embedded in the main tree after chloromethylfluorescein diacetate (Invitrogen) was used for standardisation of the branch lengths. labelling. The published genome sequences of several dictyostelid species were retrieved from dictyBase (http://dictybase.org/), NCBI (www.ncbi.nlm.nih.gov/genomes/), and SACGB 2.7 Observation (http://sacgb.fli-leibniz.de/cgi/index.pl). Genome sequences of D. polycarpum, D. deminutivum, D. polycephalum strain A Zeiss LSM510 confocal microscope was used for obser- MY1–1 and Acytostelium ellipticum were kindly provided vation of developing structures (Plan-Apochromat 10x/0.45 by Dr. Pauline Schaap. The genome sequences of D. or Plan-Apochromat 20x/0.75). For individual cells, an rosarium WS689 (strain originally provided by Dr. David Olympus FV3000 confocal microscope was used (PLA- Waddell) and D. caveatum B4–3 (strain from Dicty Stock PON60XO, NA1.42, UPLSAPO40X2, NA0.95). For GFP, Center) were constructed by the authors (unpublished), and the 488 nm line of the argon laser was used with a those of D. polycephalum strain 1JNd1 (isolated locally by 500*530 nm emission filter. For tetramethylrhodamine, the K. I.) was provided by Dr. H. Kuwayama (unpublished). HeNe 543 nm laser was used with a 560 nm long-pass emission filter.

2.5 Transformation 2.8 Expression time-course Sequences of the candidate tgr genes were amplified by PCR and inserted at the KpnI site of pDXA-GFP2 (Levi et al. Washed D. purpureum cells harvested at their exponential 2000) using In-Fusion (TAKARA BIO INC., Kusatsu, growth phase were deposited on cellulose acetate mem- Japan) or SLiCE (Okegawa and Motohashi 2015). For Dp- branes (Millipore) that had been thoroughly washed in csaA, the DNA sequence corresponding to amino acids 1 to boiled water, and the membranes were placed on agar plate 487 was fused to the DNA sequence encoding the trans- after removal of excessive water. The membranes were membrane domain (amino acid 470 to 544) of D. discoideum sampled at 3-hour intervals from 0 h to 12 h into Eppendorf gppA (Barth et al. 1994) and inserted at the KpnI site of tubes, frozen in liquid nitrogen, and stored at -80°C until pDXA-GFP2. The fusion constructs were introduced into use. Total RNA was extracted using RNeasy mini kit (Qia- AX2 cells by electroporation, and transformants were gen), and cDNA was prepared using SuperScriptII Reverse selected for resistance to G418. Contrary to the report by Transcriptase (Thermo Fisher Scientific). Transcription level Gabriel et al. (1999), this fusion gene gave rise to clear of the candidate genes were quantified by real-time quanti- fluorescence on the cell membrane (see Results). tative PCR with SYBR Premix Ex TaqTM II (Tli RNaseH Plus, Takara) on a StepOnePlus cycler (ABI). D. purpureum- Ig7 was used as an internal reference. 2.6 Labelling and mixing

Exponentially growing D. purpureum cells were washed free 3. Results and Discussion of bacteria by repeated centrifugation in 20 mM potassium phosphate, pH 6.0 (PB), resuspended in electroporation 3.1 Cell sorting in aggregation streams buffer (EP, 20 mM Na/K2 phosphate pH 7.4, 50 mM sucrose, 2 mM MgSO4 and 0.2 mM CaCl2), and labelled When differently labelled D. discoideum and D. purpureum with tetramethylrhodamine-dextran (Invitrogen, Eugene, cells were mixed at a ratio of 1:1 and allowed to develop, Oregon, USA) by electroporation (0.4 mg/mL in EP, 107 they aggregated together to form common mounds, but cells/ml, 400 ll in a 4 mm gap cuvette, 4 lF, 1000 volts, subsequently two or more tips, consisting entirely of either 1028 IS Hayakawa and K Inouye species, emerged from those mounds, consistent with the form a bidirectional gene pair, and they are expressed in results of earlier studies (Raper and Thom 1941; Jack et al. concert at the onset of multicellular stage (Benabentos et al. 2008). A pair of D. discoideum and D. purpureum fruiting 2009). The tgrB1/C1 pair is also highly polymorphic and bodies rising from a common base was the typical outcome plays a central part in kin-recognition within that species (figure 1A). We found that under slow development condi- (Benabentos et al. 2009; Hirose et al. 2011; Gruenheit et al. tions at low temperature, stream formation of the two species 2017). In the D. discoideum genome, there are two other was synchronised and cells of each species tended to align structurally similar gene pairs, consisting of a tgrB-type and lengthwise within a stream (figure 1B), suggesting the a tgrC-type genes, but they are almost invariant among involvement of differential cell adhesion. independent isolates (Benabentos et al. 2009; Gruenheit et al. 2017), and do not function as a tgrB1/C1 substitute in null mutants. In order to identify molecules responsible for 3.2 Candidate molecules for species recognition cell sorting in the mixture of D. discoideum and D. pur- in D. discoideum pureum, we introduced D. purpureum genes closely related to the D. discoideum genes encoding these candidate pro- Cells at the aggregation stage are elongated in the direction teins into D. discoideum cells and examined whether they of movement, and there is distinction between their front and affect the segregation patterns. back (Shaffer 1962). Earlier studies of cell adhesion in D. discoideum using monovalent Fab antibodies distinguished end-to-end and side-by-side cell adhesions, and the specific 3.3 Candidate molecules for species recognition adhesion molecule, contact site A (gp80) was identified as in D. purpureum being responsible for the former (Beug et al. 1973). This molecule, with immunoglobulin-like folds (TIG domain), The genome of a wild strain of D. purpureum (P1Ba) was mediates homophilic cell adhesion (Siu et al. 1987). In D. sequenced on the Illumina platform, and searched for discoideum, mutant cells lacking the gene encoding contact sequences related to the contact site A and tgr-family genes site A (csaA) do not show a significant difference in phe- of D. discoideum. We found one gene forming a cluster in notype, partially because another TIG-domain-containing phylogenetic trees with the D. discoideum contact site A protein TgrC1 is precociously expressed to complement the gene (csaA). The deduced amino acid sequence of the gene function of contact site A in csaA-null mutant cells (Gao indicates its similarity to contact site A (supplementary fig- et al. 1992; Wang et al. 2000). TgrC1 mediates cell adhesion ure 1D), including the presence of a predicted globular by heterophilic interaction with another TIG-domain-con- domain containing immunoglobulin-like folds, a putative taining protein TgrB1 (Chen et al. 2013). Although similar GPI-modification site and the C-terminal signal sequence, as to each other, they have distinct sequence features, which we well as an octapeptide similar to, but distinct from, the one in call the tgrB-type and tgrC-type (supplementary figure 1). D. discoideum contact site A which mediates homophilic On the D. discoideum genome, the tgrB1 and tgrC1 genes interaction (Siu et al. 1987; Kamboj et al. 1989). We assume

Figure 1. (A) Fruiting bodies of D. purpureum (dark-coloured sorus) and D. discoideum (pale sorus) sharing a common base. (B) Streams of aggregating cells formed from a 1:1 mixture of GFP-expressing D. discoideum cells (green) and TRITC-dextran-labelled D. purpureum cells (magenta). Cells were allowed to develop at 10°C for 24 h. The low temperature helps synchronise the formation of aggregation streams of both species. In this photograph, the cells are migrating towards the centre of aggregation located near the top right corner. Scale bars: 100 lm. Species recognition in social amoebae 1029 that this gene codes for a protein functionally equivalent to proportions and followed their development. Figure 3A D. discoideum contact site A, and refer to the gene as Dp- shows the distribution of transformant cells and D. pur- csaA. pureum cells in migrating slugs. Only the combination of We identified three pairs of tiger family genes in the P1Ba Dd(Dp-tgr2752C) and D. purpureum yielded chimaeric genome. They are composed of a tgrB-type gene and a tgrC- slugs (figure 3B). There are two types of cell distribution in type gene orienting outwards in a similar manner to the tgr this combination, one showing scattered distribution and the pairs of D. discoideum and possibly encode pairs of TIG- other showing clustering (left and middle in the middle row domain-containing proteins engaged in heterophilic inter- of figure 3A). During aggregation, cells of both species are actions (supplementary figure 1A and B). The overall simi- intermingled (figure 3A, labelled ‘streams’), but over the larity between any of the D. purpureum and D. discoideum course of slug formation, cells of the same species tended to tgr pair genes was not higher than those between two within cluster together, making the latter distribution pattern pre- either species, and the phylogenetic tree shows that the three dominant. This may be due to inherently strong cell cohe- tgr pairs are paralogous within the two species (supple- sion within each species compared to the binding solely mentary figure 1C). We call these genes Dp-tgr2752B, Dp- mediated by the introduced molecules, or there may be as yet tgr2752C, Dp-tgr11238B, Dp-tgr11238C, Dp-tgr11392B, unknown mechanisms to exclude isolated cells of other and Dp-tgr11392C depending on whether they are tgrB-like species from homogeneous cell clusters. In the other com- or tgrC-like (numbers represent contig numbers). Their binations, including the one with Dp-csaA, almost all slugs deduced amino acid sequences show moderate similarity to contained cells of either species alone (figure 3B; supple- each other (identity 30–55%) and to the Tgr proteins of D. mentary figure 2). Later in development, chimaeric slugs discoideum (identity 27–31%). formed fruiting bodies of which the sorus contained viable spores of both species (figure 3A, labelled ‘sorus’). We also observed fruiting bodies with two sori, the one at the apical 3.4 Identification of species recognition molecules end (and the stalk) almost entirely composed of D. pur- in D. purpureum pureum cells and the lower one mostly of D. discoideum cells (supplementary figure 3). When mixed with D. dis- We cloned these seven genes and introduced each of them coideum cells, all the transformants expressing the D. pur- into D. discoideum cells as a GFP-fusion construct under the pureum genes formed mixed slugs (figure 3A lower left; control of the act15 constitutive promoter. Cells of the supplementary figure 4), indicating that the molecules transformants, hereafter denoted by Dd(Dp-gene-name), derived from D. purpureum did not elicit rejection by D. showed clear GFP signal on the cell membrane as well as discoideum cells. inner membrane structures (figure 2). To see the effects of The developmental time-courses of Dp-tgr2752B and Dp- the introduced genes on the sorting behaviour, we mixed tgr2752C expression were almost identical, with peaks at the cells of each transformant with D. purpureum cells in equal mound stage, whereas the mRNA levels of the other tgr

Figure 2. Membrane localisation of the GFP-fusion products of the Dp-csaA and Dp-tgr genes in transformed D. discoideum cells. Scale bar: 10 lm. 1030 IS Hayakawa and K Inouye

Figure 3. (A) Distribution of transformed D. discoideum cells (green) and D. purpureum cells or untransformed D. discoideum cells (magenta) in slugs, sorus of a fruiting body, or aggregation streams. D. discoideum cells transformed with a D. purpureum gene as indicated in green letters were mixed with rhodamine-labelled D. purpureum or untransformed D. discoideum cells in equal proportions and incubated at 10°C (streams) or 17°C (others) for 24 h. The colours of the letters correspond to the cell labels. Same magnification except the image of the sorus. Scale bars are 100 lm for slugs and streams, 50 lm for the sorus. Results of other combinations are shown in supplementary figures 2 and 4. (B) Fraction of chimaeric slugs formed from mixtures of D. discoideum transformants and D. purpureum cells. Means and S.D. of 6 (Dp-tgr2752C)to2(Dp-tgr11238C) independent experiments are shown (total number of slugs examined 780). In the mixtures of transformed and untransformed D. discoideum cells, all the slugs contained both cells (supplementary figure 4). genes were very low (supplementary figure 5). The expres- remained mixed with D. purpureum cells throughout mul- sion of Dp-csaA showed a pattern similar to Dp-tgr2752B ticellular development, suggesting that these two genes have and Dp-tgr2752C with a lower peak. These results are distinct functions. It has been proposed that in D. discoideum consistent with the notion that the Dp-tgr2752B and Dp- the two Tgr molecules have distinct and cooperative func- tgr2752C gene products function cooperatively at the mound tions in cell recognition, with TgrC1 acting as a ligand and stage, the time when cells of different species segregate from TgrB1 its receptor (Hirose et al. 2017). The similarities each other in mixed culture. However, it was only Dd(Dp- between the Tgr pairs of the two species suggest that tgr2752C) cells, but not Dd(Dp-tgr2752B) cells, that this may also be the case in D. purpureum. The failure of Species recognition in social amoebae 1031

Figure 4. (A) Sequence polymorphism of Dp-tgr2752B and Dp-tgr2752C. Codon-wise alignments of all wild isolates were constructed for each gene, and the number of observed synonymous and non-synonymous nucleotide differences at each codon were calculated for all pairs of isolates. The graphs show synonymous differences (above the abscissa) and non-synonymous differences (below the abscissa) at each position averaged over all combinations of isolates. Domain structure predictions of the genes are shown below the graphs. The direction and scale correspond to the graphs. SP: signal peptide, TM: transmembrane region, shaded areas: TIG domains (E-values less than 1). (B) Mean number of synonymous (dark) and non-synonymous (light) substitutions of the six tgr genes. (C) Distribution of D. purpureum strain P1Ba (top row, green) or Dd(Dp-tgr2752C) (bottom row, green) and other D. purpureum wild strains (magenta) in mixed slugs. Scale bar: 100 lm.

Dd(Dp-tgr2752B) cells to mix with D. purpureum cells may figure 6). However, in D. discoideum also, tgrB1-null cells, be because Dp-Tgr2752B molecule does not function in D. which have tgrC1 but no tgrB1, mix with parental D. dis- discoideum cells due to the difference in the molecule’s coideum cells whereas tgrC1-null cells do not (Benabentos cytoplasmic tail between the two species (supplementary et al. 2009), suggesting the presence of pleiotropic roles of 1032 IS Hayakawa and K Inouye TgrC1, which were later shown to be the case (Wang and Table 1. The number of potential tgrB/C-type pairs in the dic- Shaulsky 2015). In D. purpureum also, Dp-Tgr2752C may tyostelid genomes have additional important functions in later development. Number of potential Clade Species Strain tgrB/C pairs 3.5 DNA sequence polymorphism in D. purpureum tgr Group 4 D. purpureum P1Ba 3 genes D. purpureum QSAX1 4 D. rosarium WS689 3 D. citrinum OH494 1 To test if the effects of Dp-tgr2752C are relevant to kin D. discoideum AX4 5 recognition, we examined the D. purpureum tgr genes in D. intermedium PJ-11 2 wild isolates that have SSU rDNA sequences identical to that D. firmibasis TNS-C- 5 of the D. purpureum strain P1Ba used in the present study 0014 (supplementary figure 7). Although collected at scattered Violaceum P. violaceum QSvi11 0 Group 3 D. lacteum 0 locations (inter-site distance 6.2 km to 561 km, supple- D. caveatum B4–3 0 mentary table 1), they belong to the same mating group as Polycephalum D. polycephalum MY1–1 0 confirmed by the formation of macrocysts (structures formed D. polycephalum 1JNd1 0 in the sexual reproduction cycle in social amoebae) in Group 2B P. pallidum PN500 0 pairwise mixes (supplementary table 2). The Dp-tgr2752B A. ellipticum AE2 0 Group2A A. subglobosum LB-1 0 and Dp-tgr2752C genes were both highly polymorphic Polycarpum D. polycarpum Ohio 0 (figure 4A), with the level of polymorphism being substan- Group 1 D. fasciculatum SH3 0 tially higher than those of the other tgr genes (figure 4B). In D. deminutivum M19A 0 pairwise mixed cultures under conditions favouring asexual development, i.e. the pathway leading to fruiting body for- Predicted genes showing homology to tgrB1 or tgrC1 of D. discoideum with an E-value threshold of 1e-10 were provisionally called tgrB-type mation, cells of different isolates aggregated together but gene or tgrC-type gene, respectively, and pairs of tgrB-type and tgrC- then segregated from each other to form separate fruiting type genes that show a bidirectional arrangement with an intergenic bodies (figure 4C). Notably, Dd(Dp-tgr2752C) did not form distance shorter than 1.5 kb were scored. The numbers include pairs chimaeras with D. purpureum isolates other than the one the with pseudogenes. In the species not belonging to group 4, tblastn gene was cloned from (figure 4C), indicating that Dp- searches even with much more relaxed criteria (E-value threshold of 10, intergenic distance threshold of 5 kb) did not yield anything like tgrB/ tgr2752C plays a key role in the clone-level recognition and C-type pairs (supplementary table 3) sorting in D. purpureum as well. The above results also indicate that sexual compatibility is uncorrelated with asexual compatibility. Such independency has been studies (Parikh et al. 2010; Glo¨ckner et al. 2016), and the demonstrated in natural populations of D. purpureum and D. mechanism of species recognition is also likely conserved giganteum (Sathe et al. 2013). In D. discoideum genome, between the two species. Conceivably, the kin recognition these loci are on different chromosomes (Bloomfield et al. systems of D. discoideum described by Hirose and others 2010). (Hirose et al. 2011) and that of D. purpureum are in fact the species recognition system that has attained a level of pre- cision capable of discriminating between clones within the 3.6 Evolution of species recognition species. On the other hand, the formation of chimaeric fruiting bodies by genetically heterogenous cells has also These results, taken together, suggest that Dp-tgr2752C is a been reported (Filosa 1962; Buss 1982; Strassmann et al. component of the species recognition system, which shares a 2000; Fortunato et al. 2003; Kaushik and Nanjundiah 2003; common molecular basis with the kin recognition system in Sathe et al. 2013). A possible explanation would be that the D. discoideum (Hirose et al. 2011). Considering the large cells in these studies had compatible tgr genes as a result of distance between the two species, which have followed recombination. These cells were in most cases from the same separate evolutionary paths for over a few hundred million or nearby sites, and the opportunity for sexual reproduction years (Fiz-Palacios et al. 2013; Schilde et al. 2016), it is could be high, as was shown for natural populations of D. remarkable that the introduction of a single gene from one to discoideum (Flowers et al. 2010). However, the occurrence the other made their tissues compatible with each other and of interspecific chimaeras (Sathe et al. 2013) implies the allowed coordinated morphogenesis and normal cell differ- involvement of factors not considered in the present study. entiation within the chimaeric tissues. This provides evi- To gain perspective on how the recognition systems in dence that the set of molecules involved in cell interactions social amoebae were established and maintained in evolu- in development is highly conserved between these species, tion, we searched the available dictyostelid genome as would be suggested by the comparative trascriptome sequences for such genes, and found that tgr gene pairs with Species recognition in social amoebae 1033 gene arrangements similar to tgrB1/tgrC1 of D. discoideum 1995), which is produced by prespore cells and degraded by are present only in the genomes of the clade ‘group 4’ prestalk cells (Kay and Thompson 2001). Prestalk cells use (table 1, supplementary table 3). This clade is composed of their resources to help the prespore cells migrate further and species, such as D. discoideum and D. purpureum, that use serve to make the fruiting body larger and more rigid. This cAMP as the chemotactic agent for aggregation. They typ- type of system is particularly susceptible to destabilisation ically produce large and robust fruiting bodies with a stout by invading clones which, for instance, produce larger stalk, a trait considered to be adaptive to cool and dry amount of DIF-1 while being less sensitive to it, thus gaining environments (Cavender 2013). Studies on the distribution larger than fair share of spores (Uchinomiya and Iwasa of dictyostelid slime moulds have shown that group 4 spe- 2013). Therefore, acquisition of such a ‘worker’ cell type cies are indeed predominant and abundant in areas of such could substantially increase the risk of exploitation by environments (Cavender 1973; Benson and Mahoney 1977; ‘cheater’ cells, requiring a more stringent mechanism to Cavender 1978; Swanson et al. 1999). Species of this clade prevent mixing with non-self-cells for stability. Although alone produce prestalk cells, precursor of stalk cells, whereas nothing is known about the mechanism of species recogni- outside of group 4, stalk cells differentiate from dediffer- tion outside group 4, our results of inter-species mixing entiated prespore cells at the apical end of the sorogen experiments suggest that cells are more tolerant to geneti- (Schilde et al. 2014). In D. discoideum, and presumably in cally distinct cells in those phylogenetic groups (figure 5). other group 4 species also, a sizable fraction of prestalk cells Our current hypothesis is that this cell recognition mecha- located in the subapical region (prestalk-O cells) are induced nism mediated by the polymorphic Tgr molecules was earlier in development by a diffusible molecule DIF-1 established in a group 4 ancestor and served as a prerequisite (Thompson and Kay 2000; Kay et al. 1992: van Es et al. for the invention of prestalk cells. It is conceivable that

Figure 5. Examples of between-species mixing experiments. In one-to-one mixes of cells outside group 4, genetically distinct species/ strains often remain mixed during fruiting body formation. (A) Acytostelium subglobosum (green) and A. amazonicum (magenta) in group 2A (genetic distance: 0.1321), (B) Polysphondylium filamentosum (green) and P. pallidum (magenta) in group 2B (0.0604), (C) D. polycephalum strain 2A2j (green) and D. polycephalum strain 1JNd1 (magenta) in clade polycephalum (0.0225). (D) D. lacteum (green) and D. menorah (magenta) in group 3 (0.0760), (E) D. lacteum (green) and D. rhizopodium (magenta) in group 3 (0.1529), (F) P. violaceum strain no. 6 (green) and D. laterosorum (unlabelled) in clade violaceum (0.0171). (G) P. violaceum strain P4Ba (green) and P. violaceum strain 1MN8b (magenta) in clade violaceum (0.0197), and (H) D. giganteum (green) and D. purpureum (magenta) in group 4 (0.0968). The structures shown are (part of) fruiting bodies in (B, G) and the one near the upper left corner of (H), slugs emerging from a cell aggregate in (D and E), and slugs in (A, C, F). Pairwise genetic distances (base substitutions per site) were calculated using the SSU rDNA sequences and shown in parentheses as a measure of phylogenetic distance. Only in (H) are the two species segregated completely, while the two species/strains in the other combinations remain partially to fully mixed. The genetic distance between any two of the isolates of D. purpureum used in this study was 0, nevertheless they largely segregate from each other in mixing experiments (figure 4c). On the other hand, while the genetic distance between D. purpureum P1Ba and D. discoideum is 0.1151, they could be made to remain mixed by the introduction of one gene (figure 3). Scale bars: a 25 lm, h 100 lm, others 50 lm. 1034 IS Hayakawa and K Inouye concurrent evolution of a precise allorecognition system Barth A, Mu¨ller-Taubenberger A, Taranto P and Gerisch G 1994 based on rapidly-evolving polymorphic cell adhesion Replacement of the phospholipid-anchor in the contact site A molecules and the new dedicated cell-type in the group 4 glycoprotein of D. discoideum by a transmembrane region does ancestor enabled its descendant to make robust fruiting not impede cell adhesion but reduces residence time on the cell bodies, which might have contributed to the expansion of surface. J. Cell Biol. 124 205–215 their habitats. Benabentos R, Hirose S, Sucgang R, Curk T, Katoh M, Ostrowski EA, Strassmann JE, Queller DC, Zupan B, Shaulsky G and Kuspa The contiguous arrangement of genes for highly poly- A 2009 Polymorphic members of the lag gene family mediate kin morphic complementary (key and keyhole) molecules are discrimination in Dictyostelium. Curr. Biol. 19 567–572 also associated with allorecognition in other organisms, such Benson MR and Mahoney DP 1977 The distribution of dictyostelid as ascidians (Harada et al. 2008) and plants (Takuno et al. slime molds in southern California with taxonomic notes on 2007). Compared to the complex gene organisations in these selected species. Am. J. Bot. 64 496–503 organisms, the simple gene arrangement in social amoebae Beug H, Katz FE and Gerisch G 1973 Dynamics in antigenic makes them an ideal eukaryotic model to study the evolution membrane sites relating to cell aggregation in Dictyostelium of self/non-self recognition. Our results indicate that such discoideum. J. Cell Biol. 56 647–658 studies are experimentally feasible. Bonner JT and Adams MS 1958 Cell mixtures of different species and strains of cellular slime moulds. J. Embryol. Exp. Morphol. 6 346–356 Acknowledgements Buchmann K 2014 Evolution of innate immunity: clues from invertebrates via fish to mammals. Front. Immunol. 5 1–8 Buss LW 1982 Somatic cell parasitism and the evolution of somatic We would like to thank T. Oyama and M. Ichikawa for tissue compatibility. Proc. Natl. Acad. Sci. USA 79 5337–5341 their continuous help and encouragement during the course Cavender JC 1973 Geographical distribution of Acrasiae. Mycolo- of the study. We also thank P. Schaap and H. Kuwayama gia 65 1044–1054 for genome sequence data, T. Hayakawa for advice on de Cavender JC 1978 Cellular slime molds in tundra forest soils of novo assembly of NGS data, S. Ito for discussions and help Alaska including a new species, Dictyostelium septentrionalis. in quantitative PCR, T. Shimada for the use of a qPCR Can. J. Bot. 57 1326–1332 cycler, S. Nonaka, Y. Kamei and Olympus Corporation for Cavender JC 2013 A global overview of Dictyostelid ecology with use of confocal microscopes in Department of Physics, Y. special emphasis in North American forest; in – Yamaguchi for help in interspecies mixing experiments of evolution, genomics and cell biology (eds) M Romeralo, S Baldauf non-group 4 species, M. M. Hayakawa for insightful and R Escalante (Berlin, Heidelberg: Springer-Verlag) pp149–166 comments and suggestions, and V. Nanjundiah for discus- Chen G, Wang J, Xu X, Wu X, Piao R and Siu CH 2013 tgrC1 sions and invaluable comments on the manuscript. Some of mediates cell-cell adhesion by interacting with tgrB1 via mutual IPT/TIG domains during development of Dictyostelium dis- the wild isolates were obtained from National BioResource coideum. Biochem. J. 452 259–69 Project Nenkin (http://nenkin.nbrp.jp). This work was de Wit RJW and Konijn TM 1983 Identification of the acrasin of supported by Grant-in-Aid for JSPS Fellows 12J01422 to Dictyostelium minutum as a derivative of folic acid. Cell Differ. ISH. The DNA sequences for Dp-csaA and Dp-tgrs have 12 205–210 been deposited at DDBJ/ENA/GenBank under accession Eisenberg RM 1976 Two-dimensional microdistribution of cellular numbers MK002463 to MK002469, for rDNA-ITS of the slime molds in forest soil. Ecology 57 380–384. D. purpureum strains, MK002470 to MK002481, and Filosa MF 1962 Heterocytosis in cellular slime molds. Am. assembled contigs of the genome of D. purpureum strain Naturalist 46 79–92. P1Ba, RBCH00000000. The strains used in this study can Fiz-Palacios O, Romeralo M, Ahmadzadeh A, Weststrand S, be obtained from National BioResource Project Nenkin Ahlberg PE and Baldauf S 2013 Did terrestrial diversification of (http://nenkin.nbrp.jp). amoebas (amoebozoa) occur in synchrony with land plants? PLoS One 8 e74374 Gabriel D, Hacker U, Kohler J, Mu¨ller-Taubenberger A, Schwartz JM, Westphal M and Gerisch G 1999 The contractile vacuole network of Dictyostelium as a distinct organelle: its dynamics References visualized by a GFP marker protein. Correction: 112:U3(1999). J. Cell Sci. 112 3995–4005 Asghar A, Groth M, Siol O, Gaube F, Enzensperger C, Glo¨ckner G Gao EN, Shier P and Siu CH 1992 Purification and partial and Winckler T 2012 Developmental gene regulation by an characterization of a cell adhesion molecule (gp150) involved in ancient intercellular communication system in social amoebae. postaggregation stage cell-cell binding in Dictyostelium dis- Protist 163 25–37 coideum. J. Biol. Chem. 267 9409–9415 Atzmony D, Zahavi A and Nanjundiah V 1997 Altruistic behaviour Glo¨ckner G, Lawal HM, Felder M, Singh R, Singer G, Weijer CJ in Dictyostelium discoideum explained on the basis of individual and Schaap P 2016 The multicellularity genes of dictyostelid selection. Curr. Sci. 72 142–145 social amoebae. Nat. Comm. 7 12085 Species recognition in social amoebae 1035 Gruenheit N, Parkinson K, Stewart B, Howie JA, Wolf JB and the social amoeba Dictyostelium purpureum. Evolution 63 Thompson CRL 2017 A polychromatic ‘greenbeard’ locus 542–548 determines patterns of cooperation in a social amoeba. Nat. Nanjundiah V and Sathe S 2011 Social selection and the evolution Comm. 8 14171 of cooperative groups: the example of the cellular slime moulds. Hagiwara H, Kawakami S and Hwang JY 2004 Mating system and Integr. Biol. 3, 329–342 morphology of the temperate form of Dictyostelium purpureum Nei M and Gojobori T 1986 Simple methods for estimating the Olive. Bull. Natl. Sci. Mus. Tokyo, Ser. B 30 71–78 numbers of synonymous and nonsynonymous nucleotide sub- Harada Y, Takagaki Y, Sunagawa M, Saito T, Yamada L, Taniguchi stitutions. Mol. Biol. Evol. 3 418–426 H, Shoguchi E and Sawada H 2008 Mechanism of self-sterility Nydam ML and De Tomaso AW 2011 Creation and maintenance of in a hermaphroditic chordate. Science 320 548–550 variation in allorecognition loci: molecular analysis in various Hirose S, Benabentos R, Ho HI, Kuspa A and Shaulsky G 2011 model systems. Front. Immunol. 2 79 Self-recognition in social amoebae is mediated by allelic pairs of Okegawa Y and Motohashi K 2015 A simple and ultra-low cost tiger genes. Science 333 467–470 homemade seamless ligation cloning extract (SLiCE) as an Hirose S, Chen G, Kuspa A and Shaulsky G 2017 The polymorphic alternative to a commercially available seamless DNA cloning proteins tgrB1 and tgrC1 function as a ligand-receptor pair in kit. Biochem. Biophys. Rep. 4 148–151 Dictyostelium allorecognition. J. Cell Sci. 130 4002–4012 Ostrowski EA, Katoh M, Shaulsky G, Queller DC and Strassmann Jack CN, Ridgeway JG, Mehdiabadi NJ, Jones EI, Edwards TA, JE 2008 Kin discrimination increases with genetic distance in a Queller DC and Strassmann JE 2008 Segregate or cooperate – a social amoeba. PLoS Biol. 6 e287 study of the interaction between two species of Dictyostelium. Parikh A, Miranda ER, Katoh-Kurasawa M, Fuller D, Rot G, Zagar BMC Evol. Biol. 8 293 L, Curk T, Sucgang R, Chen R, Zupan B, Loomis WF, Kuspa A Kamboj RK, Gariepy J and Siu CH 1989 Identification of an and Shaulsky G 2010 Conserved developmental transcriptomes octapeptide involved in homophilic interaction of the cell in evolutionarily divergent species. Genome Biol. 11 R35 adhesion molecule gp80 of Dictyostelium discoideum. Cell 59 Raper KB and Thom C 1941 Interspecific mixtures in the 615–625 Dictyosteliaceae. Am. J. Bot. 28 69–78 Katoh K and Standley DM 2013 MAFFT multiple sequence Romeralo M, Cavender JC, Landolt JC, Stephenson SL and alignment software version 7: improvements in performance and Baldauf SL 2011 An expanded phylogeny of social amoebas usability. Mol. Biol. Evol. 30 772–780. (Dictyostelia) shows increasing diversity and new morpholog- Katoh K, Kuma K, Toh H and Miyata T 2005 MAFFT version 5: ical patterns. BMC Evol Biol. 11 84 improvement in accuracy of multiple sequence alignment. Sakaguchi M, Murakami H and Suzaki T 2001 Involvement of a Nucleic Acids Res. 33 511–518 40-kDa glycoprotein in food recognition, prey capture, and Kaushik S and Nanjundiah V 2003 Evolutionary questions raised induction of phagocytosis in the protozoon Actinophrys sol. by cellular development. Proc. Indian Natl. Sci. Protist 152 33–41 Acad. B 69 825–852 Sathe S, Kaushik S, Lalremruata A, Aggarwal RK, Cavender JC Kay RR and Thompson CRL 2001 Cross-induction of cell types in and Nanjundiah V 2010 Genetic heterogeneity in wild isolates of Dictyostelium: evidence that DIF-1 is made by prespore cells. cellular slime mold social groups. Microb. Ecol. 60 137–148 Development 128 4959–4966 Sathe S, Khetan N and Nanjundiah V 2013 Interspecies and Kay RR, Taylor GW, Jermyn KA and Traynor D 1992 Chlorine- intraspecies interactions in social amoebae. J. Evol. Biol. 27 containing compounds produced during Dictyostelium develop- 349–62 ment - Detection by labelling with 36Cl. Biochem. J. 281 Schilde C, Skiba A and Schaap P 2014 Evolutionary reconstruction 155–161 of pattern formation in 98 Dictyostelium species reveals that Konijn TM, Barkley DS, Chang YY and Bonner JT 1968 Cyclic cell-type specialization by lateral inhibition is a derived trait. AMP: a naturally occurring acrasin in the cellular slime molds. EvoDevo 5 34 Am. Naturalist 102 225–233 Schilde C, Lawal HM, Noegel AA, Eichinger L, Schaap P and Korber B. 2000 HIV Signature and Sequence Variation Analysis; in Glo¨ckner G 2016 A set of genes conserved in sequence and Computational Analysis of HIV Molecular Sequences (eds) AG expression traces back the establishment of multicellularity in Rodrigo and GH Learn (Dordrecht, Netherlands: Kluwer social amoebae. BMC Genomics 17 871 Academic Publishers) pp 55–72 Seike T, Nakamura T and Shimoda C 2015 Molecular coevolution Levi S, Polyakov M and Egelhoff TT 2000 Green fluorescent of a sex pheromone and its receptor triggers reproductive protein and epitope tag fusion vectors for Dictyostelium isolation in Schizosaccharomyces pombe. Proc. Natl. Acad. Sci. discoideum. Plasmid 44 231–238 USA 112 4405–4410 Litman GW, Cannon JP and Dishaw LJ 2005 Reconstructing Shaffer BM 1962 The Acrasina. Adv. Morphogenesis 2 109–182 immune phylogeny: new perspectives. Nat. Rev. Immunol. 5 Shimomura O, Suthers HLB and Bonner JT 1982 Chemical identity 866–879 of the acrasin of the cellular slime mold Polysphondylium Mehdiabadi NJ, Jack CN, Farnham TT, Platt TG, Kalla SE, violaceum. Proc. Natl. Acad. Sci. USA 79 7376–7379 Shaulsky G, Queller DC and Strassmann JE 2006 Kin prefer- Shin SH, Kim S, Kim JY, Lee SJ, Um YS, Oh MK, Kim YR, Lee ence in a social microbe. Nature 442 881–882 JW and Yang KS 2012 Complete genome sequence of Mehdiabadi NJ, Kronforst MR, Queller DC and Strassmann JE Enterobacter aerogenes KCTC 2190. J. Bacteriol. 194 2009 Phylogeny, reproductive isolation and kin recognition in 2373–2374 1036 IS Hayakawa and K Inouye Siu CH, Cho A and Choi HC 1987 The contact site a glycoprotein van Es S, Hodgkinson S, Schaap P and Kay RR 1995 Metabolic mediates cell-cell adhesion by homophilic binding in Dic- pathways for differentiation-inducing factor-1 and their regula- tyostelium discoideum. J. Cell Biol. 105 2523–2533 tion are conserved between closely related Dictyostelium Stamatakis A. 2014 RAxML version 8: a tool for phylogenetic species, but not between distant members of the family. analysis and post-analysis of large phylogenies. Bioinformatics Differentiation 58 95–100 30 1312–1313 van Haastert PJM, de Wit RJW, Grijpma Y and Konijn TM 1982 Strassmann JE, Zhu Y and Queller DC 2000 Altruism and social Identification of a pterin as the acrasin of the cellular slime mold cheating in the social amoeba Dictyostelium discoideum. Nature Dictyostelium lacteum. Proc. Natl. Acad. Sci. USA 79 408 965–967 6270–6274 Strassmann JE and Queller DC 2011 Evolution of cooperation and Wang J, Hou LS, Awrey D, Loomis WF, Firtel RA and Siu CH control of cheating in a social microbe. Proc. Natl. Acad. Sci. 2000 The membrane glycoprotein gp150 is encoded by the USA. 108 10855–10862 lagC gene and mediates cell-cell adhesion by heterophilic Sucgang R, Kuo A, Tian X, Salerno W, Parikh A, Feasley CL, binding during Dictyostelium development. Dev. Biol. 227 Dalin E, Tu H, et al. 2011 Comparative genomics of the social 734–745 amoebae Dictyostelium discoideum and Dictyostelium pur- Wang Y and Shaulsky G 2015 TgrC1 has distinct functions in pureum. Genome Biol. 12 R20 Dictyostelium development and allorecognition. PLoS One 10 Swanson AR, Vadell EM and Cavender JC 1999 Global distribu- e0124270 tion of forest soil dictyostelids. J. Biogeography 26 133–148 Watts DJ and Ashworth JM 1970 Growth of myxamoebae of the Takuno S, Fujimoto R, Sugimura T, Sato K, Okamoto S, Zhang SL cellular slime mould Dictyostelium discoideum in axenic culture. and Nishio T 2007 Effects of recombination on hitchhiking Biochem. J. 119 171–174 diversity in the Brassica self-incompatibility locus complex. Zerbino DR and Birney E 2008 Velvet: algorithms for de novo Genetics 177 949–958 short read assembly using de Bruijn graphs. Genome Res. 18 Thompson CRL and Kay RR 2000 The role of DIF-1 signaling in 821–829 Dictyostelium development. Mol. Cell 6 1509–1514 Zerbino DR, McEwen GK, Margulies EH and Birney E 2009 Uchinomiya K and Iwasa Y 2013 ‘Evolution of stalk/spore ratio in Pebble and rock band: heuristic resolution of repeats and a social amoeba: cell-to-cell interaction via a signaling chemical scaffolding in the velvet short-read de novo assembler. PLoS shaped by cheating risk. J. Theor. Biol. 336 110–118 ONE 4 e8407

Corresponding editor: BJ RAO