Yellow polyketide pigment suppresses premature hatching in social amoeba

With more than 40 pks genes in the genome of D. discoideum, we had to devise a strategy to prioritize them for subsequent analyses. To this end, we grouped selected pks genes according to their transcriptional profiles (22) during the development from single cells to the multicellular fruiting body (Fig. 1A). Analysis of these data allowed us to infer developmental stage–specific functions and also to estimate relative production levels of certain polyketides. Taken together, it seems that all investigated pks genes except for pks16 were only poorly transcribed in vegetative cells, but increasingly so during development. While some were activated at early stages such as multicellular aggregation or slug formation, most of them exhibited their highest transcriptional levels at the end of the developmental cycle, when fruiting bodies are formed. The transcription profile of the gene coding for PKS5 resembled that for stlB, both of which belong to the class of pks genes transcribed in mid- to late development (Fig. 1A). They displayed a slight increase in transcription during the slug stage and peaked later during fruiting body formation. In addition, the amount of pks5 transcript even exceeded that of stlB at the end of the developmental cycle. We thus focused on pks5 because we anticipated that it may be involved in the multicellular development and/or formation of spores. The molecular mechanisms of both processes are still not fully understood.

In order to visualize the tissue- and time-specific transcription levels of pks5 (Fig. 1), we fused its promoter region with the gene encoding the red fluorescent protein (RFP) and integrated this reporter construct into an extrachromosomal expression vector. Upon introduction of this plasmid into wild-type (WT) D. discoideum AX2, we analyzed the temporospatial occurrence of fluorescently labeled cells during the transition from vegetative cells to fruiting bodies. The first accumulation of fluorescence was detected during the slug stage. In this phase, the cells were already differentiated and were found in distinct areas of the slug (Fig. 1B). Cells exhibiting pks5 promoter-driven fluorescence were located in the broad tip region with few interspersed fluorescent cells in the remainder of the slug. These slug cells colocalized with prestalk cells, in particular those that later on develop into the upper and lower cup region of the mature sorus (Fig. 1B). With progressing development, the fluorescent signals could indeed be detected in these defined regions (Fig. 1B). In mature fruiting bodies, the fluorescence remains at the tip of the sorus and in parts of the stalk, but with aging of the fruiting bodies, fluorescent cells from the upper and lower cup regions scatter throughout the whole sorus (Fig. 1B and Video S1). Since RFP expression was only detectable in few mature spores (SI Appendix, Fig. S2), we concluded that spores do not express pks5 within the sorus. Hence, the expression of pks5 is predominant in the regions of cup cells during the multicellular development of D. discoideum and since PKS5 does not contain an N-terminal leader sequence, we conclude that upper and lower cup cells of the sorus are solely responsible for the synthesis of the respective metabolite of PKS5. This finding confirmed a study in which the transcriptome of all cell types of the fruiting body of D. discoideum were examined by RNA sequencing (RNAseq), showing that the transcript of pks5 was solely detected in cup cells (23).

In order to analyze the effect of the production of the polyketide generated by pks5, a gene inactivation mutant was created using CRISPR-Cas9 editing. The required guide RNA (gRNA) had to be carefully designed, since coding regions of pks genes are extremely similar; for instance, the nucleotide sequence identity between pks5 and pks7 is approximately 88%. Sequencing of the edited region of single mutants revealed that 87% of them had alterations within their nucleotide sequence, but only 35% thereof resulted in an interruption of the open reading frame due to a frame shift and premature stop-codons (Fig. 2A). The observed mutagenesis frequency was in good agreement with published data for CRISPR-Cas9 editing in D. discoideum (24). We used three independent pks5⁻ mutants of D. discoideum AX2 exhibiting different alterations within the edited region for subsequent analyses (SI Appendix, Fig. S3).

All pks5⁻ mutants developed normally when compared with WT in terms of timing and morphology of the developmental stages, although the sori of the mutants lost most of their yellow color (Fig. 2A). To ensure that the color loss was a direct result of pks5 inactivation, we complemented pks5⁻ mutants using two different methods. First, we inserted the gene coding for PKS5 into an extrachromosomal expression vector under the control of a constitutively active actin15 promoter and subsequently transfected the pks5⁻ mutants with this expression construct, resulting in a pks5-overexpressing (pks5-oex) rescue strain (pks5⁻[pks5-oex]). Second, for a genetic reversion of the pks5⁻ mutant, the interrupted reading frame of pks5 was reconstituted by CRISPR-Cas9, yielding the pks5⁻ rescue mutant. This was done by cotransfecting a single-stranded oligonucleotide (ssOligo) with the sequence coding for PKS5 and the overlapping sequence up- and downstream of the Cas9 cutting site together with the gRNA-containing CRISPR-Cas9 vector. To distinguish the reconstituted PKS5 coding sequence from the WT sequence, we introduced silent mutations in the ssOligo, leaving the amino acid sequence of PKS5 unaffected (Fig. 2A). Whereas extrachromosomal complementation has been the method of choice for rescuing the phenotype of a knockout mutant, CRISPR-Cas9 now allows for onsite reconstitution of the disrupted gene. These two methods allow one to discern between unwanted phenotypes resulting from either polar effects or off-target mutations, respectively.

For the WT, the yellow coloration of the sori intensified over time, while the sori of the pks5⁻ mutants remained almost colorless (Fig. 2A). In addition, the sori of many fruiting bodies of pks5⁻ mutants had a glassy appearance due to the absence of the yellow pigment. The pks5⁻[pks5-oex] rescue strain had a yellow sorus; however, the basal plate and stalks also showed a yellow coloration. This can be explained by the strongly increased expression levels of pks5 compared to WT resulting from the strong, constitutively active actin15 promoter. Consequently, in the complemented strain, the coloration did not increase with the age of fruiting bodies. The reappearance of the yellow pigment and its time-dependent accumulation in the spores of the genomically reconstituted pks5⁻ rescue mutant clearly showed that the PKS5-derived metabolite is dominantly responsible for the coloring of spores of D. discoideum (Fig. 2A). Its identity as a pigment and the slight up-regulation of pks5 transcription during aggregation and the slug stage of the amoeba prompted us to compare the phototactic response of slugs of WT and the pks5⁻ mutant. Identical behavior of slugs of both strains in terms of timing and movement toward a light source revealed that the pigment does not seem to be involved in developmental phototaxis (Video S2). The highest amount of pks5 transcript in published RNAseq data (22) was detected at the end of the developmental cycle when fruiting bodies are formed, and pks5 promoter-driven fluorescence in the sorus was maintained at least for 48 h after completion of the cycle, correlating with the intensity of yellow color in aging fruiting bodies (Fig. 1 and Video S1 and Fig. 2A). We postulated that the biological function of the metabolite of PKS5 may correlate with reproduction traits, such as spore production. Although the number of spores did not differ significantly in the absence of PKS5 (SI Appendix, Fig. S4), we observed a number of empty spore coats within the sori of the pks5⁻ mutant (Fig. 2B), which may indicate disturbed spore dormancy as a result of the absence of a PKS5-derived metabolite.

To monitor metabolic differences between the pks5⁻ mutant and WT, we applied a liquid chromatography mass spectrometry (LCMS)–based comparative metabolomics approach. To this end, fruiting body extracts of both strains were subjected to high-resolution (HR-)LCMS analysis, and the differential production of metabolites was analyzed using XCMS and the CAMERA package for R (25–28). The initial significance criteria chosen were as follows: log₂ fold-change +/−4, P < 0.01. Using these criteria, a large number of metabolites were differentially regulated in the mutant (Dataset S1). In the positive detection mode, a total of 5,401 features were identified, out of which 12 were up-regulated and 297 were down-regulated, and in the negative detection mode, 1,915 features were identified with 22 up-regulated and 332 down-regulated (Fig. 3A). This indicated a shift toward a lower metabolic load and demonstrated that the full metabolic bouquet of the fruiting body requires pks5 expression. Among all down-regulated features, many were actually absent in the pks5⁻ mutant, which may indicate a broader PKS5 product palette of downstream derivatives. In a combination with a cluster analysis, in which absent features were grouped by annotated isotope peaks and adducts and then weighted by the highest mean ion intensity in WT samples, the number of possible candidates for PKS5 products was further reduced to a total of 10 (Fig. 3B). However, their low abundance in fruiting bodies as well as the overall large number of differentially regulated metabolites rendered an unambiguous identification of a direct product from PKS5 difficult.

The elution profile of most of the compounds absent in the pks5⁻ mutant pointed to a rather lipophilic class of molecules. However, the small production titers of the metabolites strongly hampered the isolation of individual promising candidates. To increase the production levels of the elusive PKS5-derived metabolites, we overexpressed the encoding gene in WT D. discoideum under the control of a constitutive actin15 promoter using an extrachromosomal expression vector. The pks5 overexpression (pks5-oex) strain indeed showed increased pigment production as fruiting bodies overall appeared even more yellow than the WT. In addition to the sori of the pks5-oex fruiting bodies, the stalks, basal plates, and surrounding media also showed a yellow pigmentation (Fig. 4A).

Following the extraction of pks5-oex vegetative cells, yellow compounds exhibiting an absorption maximum at 400 nm could be detected by high-performance LC (HPLC) analysis but were absent in the pks5⁻ mutant or the WT (Fig. 4B). HR-MS analysis revealed a series of peaks with m/z = 301.1795 [M+H]⁺, suggesting the presence of structural isomers of the same family of compounds (SI Appendix, Fig. S5). The differential metabolome analysis showed a compound with an identical mass present in the fruiting bodies of the WT but not in those of the pks5⁻ mutant (SI Appendix, Fig. S6).

In order to obtain the PKS5-derived metabolites in sufficient amounts for structure elucidation, we cultured 5 L pks5-oex strain in a stirred bioreactor and purified the main metabolites by preparative reverse-phase (RP)-HPLC (SI Appendix, Fig. S7). Structure elucidation using NMR spectroscopy revealed these compounds to be polyunsaturated long-chain fatty acids (C18) containing a hydroxyl group at C17, as well as a methyl group in either the α- or γ-position (SI Appendix, Fig. S8). The newly identified molecules were named dictyodene A (1) and B (2), respectively (Fig. 4C). The biosynthetic origin of these two compounds is consistent with a polyketide pathway, with the methyl group being introduced by a methyltransferase. Analysis of the architecture of PKS5 indeed predicts the presence of a functional methyltransferase (7). The absolute configuration of the hydroxyl group was identified as R by Mosher–ester analysis (29, 30) (SI Appendix, Fig. S9).

Differential metabolome analysis indicated the presence of minor PKS5-derived compounds that are structurally related to dictyodenes A and B. These compounds displayed similar UV absorption profiles and were produced in higher amounts in pks5-oex fruiting body extracts (SI Appendix, Fig. S10). Although the isolatable amounts of these compounds were very low, we were able to determine the structure of one compound (9) by a large-scale fruiting body cultivation. Analytical data based on NMR spectroscopy (SI Appendix, Fig. S8), tandem MS/MS (SI Appendix, Fig. S11), and Marfey’s analysis (SI Appendix, Fig. S12) agreed with a condensation product of isoleucine and dictyodene A. In addition, the presence of a methylated alanine condensation product with dictyodene A or B seems plausible based on the interpretation of MS/MS data.

Hence, we provide evidence that the dictyodenes present in the sori of D. discoideum are likely the yellow pigments, which were reported more than half a century ago. In these earlier studies, the yellow pigment was found to be enriched in ethanolic extracts of fruiting bodies and proposed to be an acidic ζ-carotenoid according to preliminary chemical analyses (31). Indeed, the published ultraviolet-visible absorption spectrum matches that of the dictyodenes with absorption maxima at λ = 396 nm and 374 nm for dictyodene B and its putative shunt product 4 (m/z = 275.1638 [M+H]⁺), respectively (SI Appendix, Fig. S13). The previous structural misassignment of these compounds may be explained by the presence of seven conjugated double bonds in both our revised structures and the presumed carotenoids. We thus provide conclusive evidence that the prominent yellow pigment within the sori of D. discoideum indeed belongs to the family of long-chain unsaturated fatty acids produced by a PKS.

Polyenes are ubiquitous in all domains of life, and their biological functions range from scavenging reactive oxygen species, light sensing, and biofilm formation to cytotoxic and antibiotic activities (32, 33). It seemed plausible that compounds that strongly absorb light may confer photoprotective properties to the amoebal spores. When analyzing the impact of UV light irradiation of fruiting bodies on spore viability, even a broad range of irradiation between 1,000 and 3,000 J/cm² did not cause any difference in spore viability between WT and the pks5⁻ mutant (SI Appendix, Fig. S14). We also did not find any major antibiotic activities when testing the purified dictyodenes against a selection of bacteria and fungi, except moderate inhibition of the basidiomycete yeast Sporobolomyces salmonicolor (SI Appendix, Fig. S15).

Because the absence of yellow sorus pigmentation was correlated with the occurrence of empty spore coats within the sori of the pks5⁻ mutant, we concluded that these coats may result from prematurely hatched spores. To assess whether hatching in the sorus was suppressed by dictyodenes, we treated WT spores with different concentrations of dictyodene B and with a mixture of all dictyodene derivatives. The mixture of dictyodene derivatives was obtained by combining all compounds with characteristic absorption at λ = 400 nm found in extracts of 4-day-old fruiting bodies of pks5-oex (SI Appendix, Fig. S10). Treatments with the purified compound and the mixture in nutrient-rich HL5 medium resulted in a dose-dependent inhibition of germination when compared to untreated spores. Notably, spores remained viable upon treatment with dictyodene B (SI Appendix, Fig. S16), supporting the function as a germination inhibitor. Complete inhibition of germination was neither achieved with dictyodene B alone nor with the mixture of all dictyodene derivatives (Fig. 4D). Germination of ∼50% of the spores was inhibited at a concentration of 58 μg/mL dictyodene B. Both metabolite 9 and the mixture of dictyodenes displayed lower inhibitory activity than dictyodene B, indicating that the latter is the dominant inhibitor within this class of molecules (SI Appendix, Fig. S17). Interestingly, on vegetative cells, dictyodene B was shown to decrease the proliferation rate with a half maximal inhibitory concentration (IC₅₀) of 4.7 μg/mL (SI Appendix, Fig. S18), although without killing the cells. Since vegetative amoebal cells are not present in the sorus of fruiting bodies, this phenomenon affects neither the physiology nor the ecology of D. discoideum.

The first documentation of the yellow pigment of D. discoideum and its developmental production profile, as well as its tight association to the spore-containing sorus, encouraged the authors to propose a reproductive function (31). Since then, several germination inhibitors of D. discoideum have been described, all of which were isolated from the spore masses. The cytokinin discadenine (3-(3-amino-3-carboxypropyl)-N6-Δ2-isopentenyladenine) is capable of inhibiting germination at concentrations of 0.14 μg/mL. Although discadenine can be chemically synthesized, the corresponding discadenine synthase has so far not been identified (34). Another germination inhibitor is N,N-dimethylguanosine; however, this compound is less potent than discadenine (35). In addition, high concentrations of ammonium phosphate within the sori of D. discoideum also inhibit spore germination in vitro (36).

Inhibition of premature germination is necessary to ensure the dispersal and eventually the survival of social amoebae. Multifactorial mechanisms of sensing environmental changes and subsequently activation of germination under favorable conditions have thus to be ensured. In the case of ammonium phosphate and dictyodenes, the underlying sensing mechanisms may be related to concentration changes induced by the swelling or shrinking of the sori according to environmental humidity, or a result of sori bursting upon contact with any object.

Many organisms have dormancy stages within their life cycles, and these are often an integral part of their survival strategies. Temporally interrupted growth and restricted mobility are usually combined with a minimization of metabolic activity and, in case of spore formation, the buildup of a rigid envelope to protect the organism against harsh environmental conditions. Eventually, dormant organisms need to recognize and respond to signals associated with favorable conditions for subsequent reawakening. For example, the binding of amino acids or sugar to a spore surface receptor of endospores of Bacillus subtilis leads to hydrolysis of the peptidoglycan content of the spore cortex, followed by vegetative outgrowth of the cell (37). In some fungi, germination inhibitors are released by the parental mycelium to ensure a broader distribution of the spores (38–40) or, in case of the plant-pathogenic fungus Fusarium solani, the inhibitor is broken down upon contact with a host cell (34). Another mechanism is known as the “scout hypothesis” or stochastic germination, where a spontaneously activated individual cell secretes a quorum-sensing compound during growth-promoting conditions to resuscitate the rest of the dormant population (41). PKS-derived autoregulatory germination inhibitors can be found in Streptomyces species (42), such as germicidin A, which is produced during the germination process; its biological function is thought to keep a proportion of the cells in the dormant state in case the environment proves to be unfavorable or to coordinate germination in higher spore densities. In Aspergillus fumigatus, the production of a germination inhibitor is induced upon contact with the spore-forming bacteria Streptomyces rapamycinicus; however, this inhibitor is probably assigned to a defense strategy of the fungus, since it only affects the bacterial germination of specific Streptomyces species (43). These examples highlight the importance of a well-regulated transition between dormancy and germination. It is thus evident that a redundancy of germination inhibitors—one of which is dictyodene B—in D. discoideum is crucial for amoebae to ensure that the right time point to hatch is reached, which eventually guarantees successful spore dissemination.

D. discoideum AX2 cells were grown in HL5 complex medium (Formedium) supplemented with 1% (wt/vol) glucose at 22 °C in cell culture dishes or shaking flasks at 140 rpm. Selection of recombinant amoebae was performed by adding 20 μg mL⁻¹ G418 (InvivoGen) to the cultivation medium. All oligonucleotides and strains used in this study are listed in SI Appendix, Tables S2 and S3, respectively. D. discoideum expression plasmids were obtained from the Dictybase Stock center (http://dictybase.org/StockCenter/StockCenter.html), and the D. discoideum all-in-one CRISPR-Cas9 vector pTM1285 was provided by NBRP Nenkin RIKEN Center for Biosystems Dynamics Research (https://nenkin.nbrp.jp).