Cell differentiation is traditionally monitored with a few marker genes, which may bias results. To understand the evolution and regulation of the spore, stalk, cup and basal disc cells in Dictyostelia, we previously performed RNAseq on purified cell-types of taxon-group representative dictyostelids. Using promoter-lacZ constructs in D. discoideum, we here investigate the spatio-temporal expression pattern of 29 cell-type specific genes. Genes selected for spore- or cup-specificity in RNAseq were validated as such by lacZ expression, but genes selected for stalk-specificity showed variable additional expression in basal disc, early cup or prestalk populations. We measured responses of 25 genes to 15 single or combined regimes of induction by stimuli known to regulate cell differentiation. The outcomes of these experiments were subjected to hierarchical clustering to identify whether common modes of regulation were correlated with specific expression patterns. The analysis identified a cluster combining the spore and cup genes, which shared upregulation by 8-bromo cyclic AMP and down-regulation by Differentiation Inducing Factor 1 (DIF-1). Most stalk-expressed genes combined into a single cluster and shared strong upregulation by cyclic di-guanylate (c-di-GMP), and synergistic upregulation by combined DIF-1 and c-di-GMP. There was no clustering of genes expressed in other soma besides the stalk, but two genes that were only expressed in the stalk did not respond to any stimuli. In contrast to current models, the study indicates the existence of a stem-cell like soma population in slugs, whose members only acquire ultimate cell fate after progressing to their terminal location during fruiting body morphogenesis.

Hierarchical clustering of gene induction data The averaged data for gene induction by stimuli, standardized either as percentage of 3 uM c-di-GMP or as fold-change relative to control (no addition) are shown in the sheet "InductionAverages" Both data matrices were used for cluster analysis using "Orange" datamining software (https://orangedatamining.com/). The columns for t0, 100nM DIF + 0.3mM cAMP and 100 nM DIF+ 8Br-cAMP, which were not tested for all genes and only in 1 or 2 experiments for others, were deleted from the matrices before analysis Distances between the effects of the treatments on the different genes were determined by Pearson correlation and a hierarchical tree was computed using average linkage (the average distance between the elements of two clusters) The full induction data matrices were re-ordered to the gene positions in the hierarchical tree in the sheet "Tree-ordered" Heatmaps of the developmental expression profiles and cell-type specificity profiles of the genes were included for presentation in Figure 3 and supplemental figure S7

Peer reviewed