Glycogen synthase kinase 3 promotes multicellular development over unicellular encystation in encysting Dictyostelia

Glycogen synthase kinase 3 (GSK3) regulates many cell fate decisions in animal development. In multicellular structures of the group 4 dictyostelid Dictyostelium discoideum, GSK3 promotes spore over stalk-like differentiation. We investigated whether, similar to other sporulation-inducing genes such as cAMP-dependent protein kinase (PKA), this role of GSK3 is derived from an ancestral role in encystation of unicellular amoebas. We deleted GSK3 in Polysphondylium pallidum, a group 2 dictyostelid which has retained encystation as an alternative survival strategy. Loss of GSK3 inhibited cytokinesis of cells in suspension, as also occurs in D. discoideum, but did not affect spore or stalk differentiation in P. pallidum. However, gsk3− amoebas entered into encystation under conditions that in wild type favour aggregation and fruiting body formation. The gsk3− cells were hypersensitive to osmolytes, which are known to promote encystation, and to cyst-inducing factors that are secreted during starvation. GSK3 was not itself regulated by these factors, but inhibited their effects. Our data show that GSK3 has a deeply conserved role in controlling cytokinesis, but not spore differentiation in Dictyostelia. Instead, in P. pallidum, one of many Dictyostelia that like their solitary ancestors can still encyst to survive starvation, GSK3 promotes multicellular development into fruiting bodies over unicellular encystment.


Background
Many unicellular protists, including Amoebozoa, survive adverse conditions by shutting down metabolism and differentiating into a walled cyst. Cysts are extremely resilient, which, in case of amoeba pathogens, prevents their eradication by immune clearance or antibiotics [1,2]. The multicellular Dictyostelia, members of Amoebozoa, evolved an additional strategy to survive starvation stress, in which amoebas aggregate to form a multicellular fruiting structure and differentiate into walled spores and stalk cells. Dictyostelia can be subdivided into four major groups, and while many species in groups 1-3 have retained encystation as an alternative survival strategy, it was lost in group 4, which contains the model organism Dictyostelium discoideum [3,4]. In D. discoideum, both secreted and intracellular cyclic AMP (cAMP) play major roles in regulating the multicellular developmental programme. Secreted cAMP, acting on G-protein-coupled cAMP receptors (cARs), acts as a chemoattractant to coordinate aggregation and morphogenesis, and additionally induces prespore differentiation, while inhibiting stalk differentiation. Intracellular cAMP, acting on PKA, triggers the maturation of spore and stalk cells and keeps spore dormant in the fruiting body. cAMP is synthesized by the adenylate cyclases ACA, ACR and ACG, but intracellular levels are critically regulated by the cAMP phosphodiesterase RegA. RegA is activated/inhibited by sensor histidine kinases/phosphatases, which are the targets for signals that control timely spore and stalk maturation and spore germination [5,6].

Open Access
EvoDevo *Correspondence: y.kawabe@dundee.ac.uk 1 School of Life Sciences, University of Dundee, MSI/WTB Complex, Dow Street, Dundee DD15EH, UK Full list of author information is available at the end of the article Comparative functional analysis of PKA, ACR, ACG and RegA in the group 2 Dictyostelid Polysphondylium pallidum and the solitary amoebozoan pathogen Acanthamoeba castellani revealed that the intracellular role of cAMP in spore and stalk maturation and spore dormancy is evolutionary derived from a second messenger role in stress-induced encystation [7][8][9][10][11]. PKA, ACR and RegA are deeply conserved in Amoebozoa, and their sequenced genomes contain many sensor histidine kinase/phosphatases, which could act as food/stress sensors, respectively, to regulate RegA [12,13].
While PKA is required for both the spore and stalk cell differentiation pathways, glycogen synthase kinase 3 (GSK3), a component of the wnt/wingless pathway that regulates many cell fate decisions in metazoa [14,15], is in D. discoideum considered to selectively promote prespore over prestalk differentiation as target for secreted cAMP, which activates GSK3 [16,17]. We are interested in expanding the range of encystation-inducing proteins that could act as therapeutic targets to prevent encystation of pathogens. We therefore investigated whether, similar to ACR, RegA and PKA, GSK3′s role in sporulation was also evolutionary derived from a role in encystation.
To address this issue we deleted the GSK3 gene of P. pallidum, which, in addition to fruiting body formation, has retained encystation as an alternative survival strategy. Surprisingly, loss of GSK3 had no negative effect on P. pallidum sporulation and promoted instead of inhibited encystation.

Growth and development
Polysphondylium pallidum (Pp), strain PN500, was routinely grown in association with Escherichia coli or Klebsiella aerogenes on lactose-peptone (LP) agar. For multicellular development, Pp cells were harvested in 20 mM K/K-phosphate, pH 6.5 (KK2), washed free from bacteria and incubated at 10 6 cells/cm 2 and 21 °C on nonnutrient agar. To determine growth rate, Pp cells were inoculated at 10 5 cells/ml in KK2 with autoclaved Klebsiella aerogenes at OD 600 = 15.

Amplification of a Pp GSK3 ortholog
The Pp GSK3 gene was amplified by PCR from genomic DNA, using redundant primers GSKredF and GSKredR (Additional file 1: Table 1), which are complementary to amino-acid sequences CHRDIKP and GTPTE/R/KQ, respectively, that are conserved in eukaryote GSK3 proteins. The PCR products were subcloned, and their DNA sequence was determined from 3 independent clones. The complete 1350-bp coding sequence of the Pp GSK3 with 3003-bp 5′ and 1579-bp 3′ UTR was obtained by inverse PCR with primer pair GSKINV1 and GSKINV2 (Additional file 1: Table 1), using religated HindIII or BglII-digested Pp gDNA as template, respectively. All PCR products were subcloned in pBluescript II KS (-) (Stratagene) or pCR4-XL-TOPO (Invitrogen) and sequenced.
To determine the nucleotide sequence of the Pp GSK3 mRNAs, polyA + RNA was isolated from Pp cells. Fulllength cDNAs were subsequently synthesized by RNAligation-mediated rapid amplification of 5′ and 3′ cDNA ends (RLM-RACE) and RT-PCR using the GeneRacer kit (Invitrogen) according to the manufacturer's instructions.

DNA constructs and transformation Vectors for GSK3 gene disruption
Partial GSK3 sequence with 2.2-kb 5′ UTR and 2.9-kb 3′ UTR was amplified by inverse PCR from EcoRI-digested and religated Pp gDNA, using primers GSKINV3 and GSKINV4 (Additional file 1: Table 1) which contain KpnI sites. The KpnI-digested PCR product was cloned into KpnI-digested pLoxNeoII∆EcoRI, which was generated from pLoxNeoII [10] by destroying its EcoRI site by digestion with EcoRI, fill-in with Klenow and self-ligation with T4 ligase. This yielded vector pPp-GSK3-KO, which was linearized by EcoRI digestion and transformed into Pp cells as described previously [18]. The gene disruption was confirmed by Southern blot analysis (Additional file 1: Fig. 1). To remove the Neo cassette, the knockout cells were transformed with pA15NLS.Cre for transient expression of Cre-recombinase [10] and G418-sensitive clones were selected.

Encystation assay
For quantification of encystation, Pp cells were grown in a suspension of autoclaved K. aerogenes in KK2, until cell proliferation reached stationary phase. Cells were washed free of bacteria, resuspended in KK2 at 10 7 cells/ml and shaken at 180 rpm and 21 °C for 48 h. Aliquots of 0.1 ml were sampled at regular intervals and supplemented with 1 µl 0.1% Calcofluor (which reacts to cellulose in the cyst wall). Total amoeba and cyst numbers were determined by counting cells in a haemocytometer under phase contrast and UV illumination, respectively. 300-500 cells were counted for each time point.

Isolation and disruption of a GSK3 homologue in Pp
To identify the role of GSK3 in Pp development, we first amplified a full-length GSK3 gene from Pp gDNA by combining PCR with degenerate primers and inverse PCR. The 2124-bp coding region contained several introns and to elucidate the gene model, we determined mRNA sequence by RT-PCR and RLM-RACE. This revealed that the GSK3 gene consists of 5 exons and 4 introns and encodes 449 amino acids. Pp GSK3 shared 92% sequence identity to D. discoideum (Dd) GSK3 and 60-70% identity with GSK3s from plants, animals and other Amoebozoa (Additional file 1: Figs. 2, 3). Query of Dictyostelid genomes with Pp GSK3 and phylogenetic inference from alignments of the closest hits shows that Pp GSK3 is orthologous to Dd GSK3.
To disrupt the GSK3 gene in Pp by homologous recombination, we transformed Pp with a construct in which the floxed A15neoR cassette is flanked by two fragments of the GSK3 gene, and obtained two gsk3 null clones from about 1000 G418-resistant clones (Additional file 1: Fig. 1). To confirm that the phenotype of the gsk3 − mutant was due to the loss of GSK3, the G418 resistance cassette of the mutant was removed by transformation with Cre-recombinase and a complementation vector, which contains Pp GSK3 inclusive of its promoter, was introduced into the disruptant.

Growth phenotype of the Pp gsk3 null mutant
The D. discoideum (Dd) gsk3 − mutant becomes multinucleate due to a cytokinesis defect, when grown in suspension in axenic medium, but not when grown on agar with bacteria as food source [20]. Pp grows poorly in axenic medium, but can be grown in suspension on autoclaved bacteria. We first tested whether Pp gsk3 − shows a similar cytokinesis defect as Dd gsk3 − . When grown in suspension with autoclaved Klebsiella aerogenes, the Pp gsk3 − cells seemed larger than wild-type cells. DAPI staining revealed that most gsk3 − cells contained multiple nuclei, while most wild-type cells contained a single nucleus (Fig. 1a). Complementation of gsk3 − with GSK3 restored the mononucleate phenotype. When the Pp gsk3 − mutant was grown with bacteria on agar, cells remained mononucleate and had the same size as wild-type cells (Fig. 1a).
Counting of the number of nuclei per cell showed that less than 5% of wild-type, gsk3 − /GSK3 or gsk3 − cells grown on agar had more than one nucleus per cells, while 72% of gsk3 − cells grown in suspension had two or more nucleo (Fig. 1b). The cytokinesis defect also reduced the doubling time of gsk3 − cells grown in suspension from 6.4 h to 4.3 h as well as the cell density reached at stationary phase (Fig. 1c). However, proliferation of gsk3 − cells on solid substratum was normal, as judged from the increase in plaque size of gsk3 − cells grown clonally on agar with bacterial lawns (Fig. 1d). Evidently, like Dd gsk3 − , Pp gsk3 − shows defective cytokinesis when grown in suspension, but not on solid substratum, indicating that the requirement of GSK3 for proper cytokinesis is conserved in Dictyostelia.

Developmental phenotype of the Pp gsk3 null mutant
Pp development differs from that of Dd in several aspects ( Fig. 2A). There is no migrating slug stage with prestalk cells as in Dd. Pp cells first all differentiate into prespore cells, only to dedifferentiate into stalk cells at the tips of primary and secondary sorogens. There are also no basal disc cells to support the stalk. However, while Dd fruiting bodies are unbranched, Pp forms regular whorls of side branches out of cell masses that pinch off from the rear of the primary sorogen. Additionally, Pp has retained the ancestral survival strategy of encystation of individual amoebas under conditions that are unfavourable for aggregation, which is lost in Dd.
The Dd gsk3 − mutant forms abnormal fruiting bodies with a large basal mass of stalk-like cells and relatively few spores [16]. However, Pp gsk3 − cells appeared to form normal fruiting bodies with a neat array of stalk cells and the multiple spore heads that are common to this species (Fig. 2B). Both the spore and stalk walls stained positively with Calcofluor White, a compound that fluoresces when interacting with cellulose, indicating that they had properly reached their mature celluloseencapsulated state (Fig. 2C). The elliptical spores of the Pp gsk3 − mutant were morphologically indistinguishable from those of wild-type cells (Fig. 2D). We determined sporulation efficiency of the gks3 − mutant by counting the number of spores differentiating from a known number of amoebas. Figure 2E shows that the Pp gsk3 − cells sporulated as efficiently as wild-type cells. Spore numbers for both exceeded that of plated amoebas, which is due to some cell division still occurring after plating the cells. Spore viability was also normal in Pp gsk3 − , since Pp gsk3 − spores germinated as efficiently as wild-type spores (Fig. 2F).
Using antispore serum, which apart from spores also detects vesicles with prefabricated spore coat components in prespore cells [20], we assessed whether Pp gsk3 − cells normally differentiate in prespore cells. Figure 3 shows that both wild-type and gsk3 − cells show the same pattern of spore antigen expression, with only the utmost tip of the sorogen and the stalk devoid of spore antigen as is the norm for this species [4] (Fig. 2a). These experiments indicate that GSK3 is not required for either prespore or spore differentiation in Pp.

Encystation of gsk3 null mutant
When, as in the experiments above, gsk3 − cells are freed from bacteria and then plated on non-nutrient agar, the greater majority of amoebas aggregate and become incorporated into fruiting bodies with similar timing as wild-type cells. However, when cells are left on the culture plate after the bacteria have been eaten, we observed that many gsk3 − cells did not aggregate and develop into fruiting bodies, but remained on the agar surface. This occurred particularly when plates were kept in the dark, where wild-type cells still developed normally, leaving very few cells behind on the agar surface (Fig. 4Aac). The prostrate gsk3 − cells (the large opaque area in Fig. 4Ad) were round and refractile (Fig. 4Ae), and Calcofluor White staining (Fig. 4Af ) revealed that they had cellulose walls and were actually microcysts.
We also observed that gsk3 − cells formed microcysts more readily after consumption of autoclaved K. aerogenes when grown in suspension (Fig. 4B). Both wildtype and gsk3 − cells reach stationary phase after about 48 h in suspension culture (Fig. 1c) with all bacteria being consumed within 60 h. Both wild-type and gsk3 − cells started to encyst around 60-72 h (Fig. 4C). After 120 h only 20% of wild-type cells had encysted as opposed to 90% of gsk3 − . Complementation of gsk3 − cells with GSK3 reduced their ability to encyst to that of wild type, indicating that loss of GSK3 potentiates encystation.

Production of and response to cyst-inducing factors
The increased ability of gsk3 − cells to encyst could either be due to gsk3 − cells producing more of an encystation-inducing factor or to being more sensitive to such a factor. To test the possibility that gsk3 − cells produce more of an encystation-inducing factor, we prepared supernatants from suspension cultures of either wild-type or gsk3 − cells at 60 h of culture, when bacterial food is just depleted. Supernatants prepared from both wild-type and gsk3 − cultures strongly induced encystation of gsk3 − cells, but were much less effective in wild-type cells. Incubation with water as control was ineffective to induce encystation of gsk3 − cells (Fig. 5a). Also phosphate buffer and medium prepared from heat-killed K. aerogenes were ineffective to induce encystation (data not shown). These results indicate that gsk3 − cells responded more strongly to encystation-inducing factor(s), rather than secreting more of such factors and that their encystation tendency is cell autonomous.
Osmolytes were previously reported to effectively induce encystation [21], with NH 4 Cl and KCl being more effective than other ions or solutes [22]. We compared the effects of a range of osmolytes on encystation in gsk3 − and wild-type cells. In the presence of 200 mM sorbitol, about 60% of wild-type cells encysted after 48 h of incubation, against 80% of gsk3 − cells (Fig. 5b). At 100 mM sorbitol, only 10% of wild-type cells encysted, but still 80% of gsk3 − cells, while 50 mM did not induce wild-type encystation anymore, but still caused 50% of gsk3 − cells to encyst (Fig. 5b). On the other hand, gsk3 − cells complemented with GSK3 and wild-type cells overexpressing GSK3 were even less responsive to sorbitol than wild-type cells. Similar results were obtained with other osmolytes. Both 10 mM of KCl and NH 4 Cl and 20 mM glucose induced 50-80% encystation of gsk3 − cells, compared to 1-13% of wild-type cells (Fig. 5c). These findings indicate that gsk3 − cells were much more sensitive to osmolytes than wild-type cells. While the naturally secreted encystation-inducing factor is unknown, NH 3 / NH 4 + is a candidate, because NH 3 is produced in large amounts by protein degradation in starving cells.
(See figure on next page.) Fig. 2 Developmental phenotype of the Pp gsk3 − mutant. A Cartoon highlighting differences between Pp and Dd development. B Fruiting bodies. Pp wild-type and gsk3 − cells were freed from bacteria and developed for 24 h on non-nutrient agar at 10 6 cells/cm 2 and photographed. Bar: 0.5 mm. C Stalk cells. The fruiting bodies were picked up, deposited in 0.001% Calcofluor on a slide glass and photographed under phase contrast (Ph) and epifluorescence (Fl) Bar: 50 µm. D Spores. Prepared as in panel B, but photographed at higher magnification. Bar: 10 µm. E Sporulation efficiency. 4 × 10 6 freshly harvested Pp amoebas were plated on 2 × 2 cm 2 nitrocellulose membranes, supported by non-nutrient agar. After completion of fruiting body formation, membranes were vortexed in 0.1% Triton-X100 and spore numbers were counted and expressed as percentage of plated cell numbers. Means and SD from 3 independent experiments. F Spore viability. Pp spores were harvested from 7-day-old fruiting bodies, treated for 10 min with 0.1% Triton-X100 to remove amoebas, counted and clonally plated on LP agar with E. coli. Emerging Pp colonies were counted after 4-5 days. Means and SD from 4 independent experiments. There were no significant differences between wild-type (WT) and gsk3 − cells in panels D and E (t test, P > 0. 5) We next measured whether NH 4 Cl affects GSK3 activity, measured in cleared lysates of wild-type cells starved in suspension buffer. Figure 5d shows that the ability of GSK3 to phosphorylate phosphoglycogen synthase peptide-2 increases up to fourfold after 24 h of starvation and up to eightfold after 48 h. However, 10 mM NH 4 Cl had no effect on GSK3 activity. The gsk3 − cells showed only 10.8 ± 1.8% of the kinase activity of wild-type cells after 24 h of starvation, indicating that non-specific phosphorylation of phosphoglycogen synthase peptide-2 in this experiment was low. This experiment shows that GSK3 is not itself regulated by NH 4 Cl, but may inhibit the pathway that mediates NH 4 Cl-induced encystation.

GSK3 has a conserved role in cytokinesis of Dictyostelia
Pp GSK3 null mutant cells were larger than those of wild type and contained multiple nuclei, indicative of a defect in cytokinesis (Fig. 1). When grown on solid substratum, these defects were not observed. This behaviour is also found in the Dd gsk3 − mutant [23] and in several Dd mutants in cytoskeletal proteins [24][25][26]. In both Dd and animals, GSK3 associates with the mitotic spindle during cell division [23,27] and has for animals been described to phosphorylate microtubule-associated proteins, with the loss of GSK3 activity causing defects in spindle alignment [27,28]. Apparently, this role of GSK3 is deeply conserved between animals and Dictyostelia. On solid substratum, Dictyostelium cells can also undergo cytokinesis by actin-based traction forces [26], a process which likely does not require GSK3.

Loss of GSK3 does not affect prespore and spore differentiation in Pp
Unlike the Dd gsk3 − mutant in strain DH1 [16], the Pp gsk3 − mutant showed no defects in prespore or spore differentiation (Figs. 2, 3). In Dd, the cAMP receptor cAR3 and the tyrosine kinases Zak1 and Zak2 are considered to mediate cAMP induction of prespore gene expression by GSK3 by phosphorylating GSK3 at tyrosine residues Y214 and Y220. Conversely, cAMP acting on cAR4 suppresses GSK3 activity by stimulating a tyrosine phosphatase [17,[29][30][31]. The Dd DH1/gsk3 − fruiting bodies consist mostly of a basal mass of stalk-like vacuolated cells and few spores, but this extreme phenotype was not observed in a GSK3 knockout in Dd strain AX2, which forms fruiting bodies with shorter stalks, but normal spores [32]. cAMP-induced prespore gene expression is normal in AX2/gsk3 − , but the mutant is hypersensitive to induction of the stalk marker gene ecmB by DIF-1, and like the DH1/gsk3 − , car3 − and zak1 − mutants, does not show cAMP inhibition of ecmB [17,29,30,32]. This suggests that GSK3 indirectly favours the spore pathway in Dd by preventing its inhibition by DIF-1. DIF-1 was originally identified as the stalk-inducing factor of Dd, but deletion of its biosynthetic pathway revealed that it was not required for the differentiation of the stalk, but of the basal disc [33][34][35]. The basal disc cells are phenotypically identical to stalk cells and also express ecmB [36]. Species outside of group 4, including Pp, do not form a basal disc and have neither cAR3 nor Zak1 or Zak2 in their genomes [10,37]. It therefore appears that the role of GSK3 in regulating prespore/basal disc proportions newly evolved in group 4.

GSK3 controls the decision between encystation and fructification in Pp
Earlier studies showed that similar to Dd [5], PKA is required for entry into multicellular development and for spore and stalk maturation in Pp, and additionally for entry into encystation [7,11], suggesting that PKA's roles in multicellular development are evolutionary derived from a more ancestral role in encystation. On the other hand, GSK3 appears to control the decision to either encyst or aggregate and form fruiting bodies. The Pp gsk3 − mutant formed cysts under conditions where wildtype cells normally aggregate and was hypersensitive to secreted factors and osmolytes that induce encystation (Figs. 4, 5). However, GSK3 activity was itself not regulated by these factors. GSK3 was also not obviously regulated at the expression level, since it is already present in feeding amoebas and modestly upregulated during both encystation and multicellular development (Additional file 1: Fig. 2). In Dd, GSK3 is required for the upregulation of about 81 genes and downregulation of 105 others in early development [38]. The hypersensitivity of Pp gsk3 − cells to encystation-inducing factors might occur if sensors for these factors or components of their signal Wild-type cells were incubated with and without 10 mM NH 4 Cl and gsk3 − cells without NH 4 Cl only. Cell extracts were prepared at the indicated time points and incubated with [γ-32 P]ATP and the GSK3 substrate phosphoglycogen synthase peptide-2 in the presence and absence of the GSK3 inhibitor LiCl. After 8 min, 32 P incorporation in the peptide was measured and non-specific 32 P incorporation in the presence of LiCl was subtracted [17]. Means and SE of 4 experiments are presented