Homeodomain-like DNA binding proteins control the haploid-to-diploid transition in Dictyostelium

Abstract

Homeodomain proteins control the developmental transition between the haploid and diploid phases in several eukaryotic lineages, but it is not known whether this regulatory mechanism reflects the ancestral condition or, instead, convergent evolution. We have characterized the mating-type locus of the amoebozoan Dictyostelium discoideum, which encodes two pairs of small proteins that determine the three mating types of this species; none of these proteins display recognizable homology to known families. We report that the nuclear magnetic resonance structures of two of them, MatA and MatB, contain helix-turn-helix folds flanked by largely disordered amino- and carboxyl-terminal tails. This fold closely resembles that of homeodomain transcription factors, and, like those proteins, MatA and MatB each bind DNA characteristically using the third helix of their folded domains. By constructing chimeric versions containing parts of MatA and MatB, we demonstrate that the carboxyl-terminal tail, not the central DNA binding motif, confers mating specificity, providing mechanistic insight into how a third mating type might have originated. Finally, we show that these homeodomain-like proteins specify zygote function: Hemizygous diploids, formed in crosses between a wild-type strain and a mat null mutant, grow and differentiate identically to haploids. We propose that Dictyostelium MatA and MatB are divergent homeodomain proteins with a conserved function in triggering the haploid-to-diploid transition that can be traced back to the last common ancestor of eukaryotes.

Fig. 1. Solution structures of MatA and MatB determined by NMR spectroscopy.

The folded core domains of MatA (A) and MatB (B) both contain three α helices arranged similarly to the homeodomain fold (helix 1 in blue, helix 2 in green, and helix 3 in red). The tail regions of MatA (C) and MatB (D) extend flexibly away from the well-folded core (N-terminal tails in pale cyan and C-terminal tails in light orange). Heteronuclear ¹⁵N{¹H} NOE data for MatA (E) and MatB (F) suggest that, in both cases, there may be some structure in the disordered tails (see text). Relationships between the orientations of the different structural views are indicated on the figure; relative scalings of the views were chosen for clarity.

Fig. 2. Sequence and structure-based alignments for MatA and MatB.

(A) Sequence alignment of MatA and MatB homologs (CLUSTAL color scheme) from a variety of Dictyostelium species shows a high degree of conservation, implying that the structure is very likely to be conserved between these species. The separate lower row shown in the core region is a structure-based alignment of MatA with S. cerevisiae MATα2, which demonstrates that many of the core hydrophobic residues (indicated with blue dots) are also conserved between these two proteins. The structures of MatA (B) and MATα2 (C) show how, in both cases, the side chains of these conserved hydrophobic residues (shown in yellow) are arranged to form the core of the structure. Side chains of solvent-exposed basic residues on the third helix that are likely (MatA) or known (MATα2) to interact with the phosphate backbone of the DNA upon binding are shown in turquoise.

Fig. 3. Assessing the DNA binding activity of MatA and MatB.

(A) EMSA experiments show that adding an increasing concentration of MatA to a dsDNA oligonucleotide causes the free DNA band to be progressively replaced by a broad smear, indicating (nonspecific) binding (see text). (B) Mutating residues Lys⁷² and Lys⁷⁶ to Ala abolishes this interaction, implicating these residues in the mechanism of DNA binding. (C) Addition of MatB causes a similar pattern to that seen for the addition of MatA. (D and E) Electrostatic potential surface of MatA, omitting the tail regions for clarity. The orientation shown in (D) (same as in Fig. 1A) shows the pronounced basic patch resulting from the conserved basic residues on the surface of helix 3. (F) Adding various lengths of dsDNA to samples of ¹⁵N-labeled MatA causes peaks to shift in the ¹⁵N-¹H HSQC (heteronuclear single-quantum coherence) NMR spectrum; these CSPs can be plotted as a histogram (G) and mapped as a color ramp onto the lowest energy structure of MatA (H), shown in the same orientation as (D). This shows that many of the shifts map to the third helix, again implicating this region in direct interactions with the DNA; Leu³², which is N-terminal to the core folded domain, is also strongly affected. The NMR experiments used MatA (20 μM) and DNA (80 μM) in 25 mM phosphate (pH 6), 50 mM NaCl, and 50 μM EDTA. ppm, parts per million.

Fig. 4. Using a macrocyst formation assay to assess the degree to which MatA/B chimeras exhibit MatA-type functionality.

(A) A series of constructs were produced by exchanging the tails and core regions of MatA and MatB. (B) Vegetative D. discoideum of different mating types, if approaching starvation under dark and humid conditions, may fuse to form a zygote giant cell. This diploid cell attracts nearby cells to form a precyst enclosed by a cellulose wall; this, in turn, matures into a dormant macrocyst, which germinates if conditions allow to release tens or hundreds of haploid progenies. (C) In this assay, chimeric constructs are defined as MatA-type if they allow cells to form macrocysts with a type II tester strain. (D) Schematic illustrating the roles of each mat gene in crosses between different mating types: Either matA or matB is required for mating compatibility with type III strains, whereas matA also confers mating with strains carrying matC (24). (E to G) Only chimeras with the C-terminal tail of MatA confer mating compatibility with type II cells, implying that this region is principally responsible for the different functions of MatA and MatB, at least within this context.

Fig. 5. The mat locus controls the growth mode of Dictyostelium diploids.

(A) Parasexual diploids were selected by mixing temperature-sensitive mutants, allowing low-frequency cell fusions at the permissive temperature, and then plating cells in cocultures with Klebsiella bacteria at the restrictive temperature. (B) In type I plus type II mixtures, no plaques of growing amoebae appeared, corroborating earlier findings (33); in contrast, a matA null strain in the same type I background readily produced plaques of temperature-tolerant diploids. (C) The contributions of other mat genes were tested by expressing green fluorescent protein (GFP)– or monomeric red fluorescent protein (mRFP)–tagged proteins in matA null strains, either using high-temperature selection for diploids in crosses with mat null, type I, and type II strains as above (for matA, matB, matC, and matS) or using double-drug selection (52). Growth and proliferation of diploids is represented by “+,” and absence of growth is represented by “−.” These results show that mat genes act to prevent growth and proliferation of sexual diploids, whether using bacteria or nutrient broth as food. (D) mat hemizygous diploids grow and develop asexually very similarly to haploids. Scale bar, 2 mm. (E) These hemizygous diploids form macrocysts when mixed with type I cells (NC4) but not with type II cells (V12M2), behaving in the same manner as type II haploid cells. Scale bar, 0.2 mm. (F) Dictyostelium mat mutants therefore display a form of homeosis in which diploid individuals behave as haploid (35): When type I cells from which matA has been deleted fuse with type II cells, the resultant diploid acts as a type II haploid and not a type I/type II zygote.

Fig. 6. Hypothetical model for the regulation of mating-type determination and the haploid-diploid transition by the Dictyostelium Mat proteins.

In the haploid type I cell, MatA binds to DNA in cooperation with (as yet unidentified) MatA-specific partner proteins via interactions involving its C-terminal tail, activating the expression of type I–specific genes. In the haploid type II cell, MatB interacts with its own distinctive partner proteins (again via its C-terminal tail) to bind DNA and activate expression of type II–specific genes. We hypothesize that MatC activates the expression of further type II–specific genes, but little is currently known about MatC to suggest how it functions. Following the fusion of type I with type II cells at the beginning of the sexual cycle, the diploid type I/type II cell now contains both MatA and MatC, and these proteins may interact via the C-terminal tail of MatA to activate the expression of diploid-specific genes. However, no such interaction occurs between MatC and the C-terminal tail of MatB in the haploid type II cell, and so, diploid-specific genes are not expressed. The Mat proteins may also cooperate to repress haploid-specific gene expression in diploid cells, whereas in haploid cells, they may be involved in repressing diploid-specific gene expression before cell fusion. In our model, we propose that MatS functions in a similar way to MatC in haploid type III cells to activate specific gene expression, and interactions between MatA/MatS and MatB/MatS pairs (similar to those between MatA/MatC in a type I/type II diploid cell) activate diploid-specific expression in type I/type III and type II/type III diploids, respectively.