Extracting transcriptional events from temporal gene expression patterns during Dictyostelium development.

MOTIVATION: The DNA microarray technology can generate a large amount of data describing the time-course of gene expression. These data, when properly interpreted, can yield a great deal of information concerning differential gene expression during development. Much current effort in bioinformatics has been devoted to the analysis of gene expression data, usually via some 'clustering analysis' on the raw data in some abstract high dimensional space. Here, we describe a method where we first 'process' the raw time-course data using a simple biologically based kinetic model of gene expression. This allows us to reduce the vast data to a few vital attributes characterizing each expression profile, e.g. the times of the onset and cessation of the expression of the developmentally regulated genes. These vital attributes can then be trivially clustered by visual inspection to reveal biologically significant effects. RESULTS: We have applied this approach to microarray expression data from samples isolated every 2 h throughout the 24 h developmental program of Dictyostelium discoideum. mRNA accumulation patterns for 50 developmental genes were found to fit the kinetic model with a p-value of 0.05 or better. Transcription of these genes appears to be initiated in bursts at well-defined periods during development, in a manner suggestive of a dependent sequence. This approach can be applied to analyses of other temporal gene expression patterns, including those of the cell cycle.

Expression patterns of cell-type-specific genes in Dictyostelium.

Cell-type specific genes were recognized by interrogating microarrays carrying Dictyostelium gene fragments with probes prepared from fractions enriched in prestalk and prespore cells. Cell-type specific accumulation of mRNA from 17 newly identified genes was confirmed by Northern analyses. DNA microarrays carrying 690 targets were used to determine expression profiles during development. The profiles were fit to a biologically based kinetic equation to extract the times of transcription onset and cessation. Although the majority of the genes that were cell-type enriched at the slug stage were first expressed as the prespore and prestalk cells sorted out in aggregates, some were found to be expressed earlier before the cells had even aggregated. These early genes may have been initially expressed in all cells and then preferentially turned over in one or the other cell type. Alternatively, cell type divergence may start soon after the initiation of development.

Requirements for the adenylyl cyclases in the development of Dictyostelium.

It has been suggested that all intracellular signaling by cAMP during development of Dictyostelium is mediated by the cAMP-dependent protein kinase, PKA, since cells carrying null mutations in the acaA gene that encodes adenylyl cyclase can develop so as to form fruiting bodies under some conditions if PKA is made constitutive by overexpressing the catalytic subunit. However, a second adenylyl cyclase encoded by acrA has recently been found that functions in a cell autonomous fashion during late development. We have found that expression of a modified acaA gene rescues acrA- mutant cells indicating that the only role played by ACR is to produce cAMP. To determine whether cells lacking both adenylyl cyclase genes can develop when PKA is constitutive we disrupted acrA in a acaA- PKA-C(over) strain. When developed at high cell densities, acrA- acaA- PKA-C(over) cells form mounds, express cell type-specific genes at reduced levels and secrete cellulose coats but do not form fruiting bodies or significant numbers of viable spores. Thus, it appears that synthesis of cAMP is required for spore differentiation in Dictyostelium even if PKA activity is high.

Spore coat formation and timely sporulation depend on cellulose in Dictyostelium.

Cellulose is a major component of the extracellular coat that surrounds the terminally-differentiated spore of Dictyostelium. It is sandwiched between two layers of proteins that derive from prespore vesicles by exocytosis. Strains unable to synthesize cellulose due to null mutations in the gene encoding the catalytic subunit of cellulose synthase (dcsA) failed to make detergent-resistant spores but produced small, highly refractile, round spore-like cells up to a day late. Although these cells resembled spores in appearance, they were unstable, only transiently ellipsoid in shape, and sensitive to hypo-osmotic shock, drying, or detergents. Differentiation of these pseudo-spores was induced in the normal time frame by activation of the cAMP-dependent protein kinase or co-development with wild type cells, and coat proteins were secreted by the dcsA-null cells at the same time as wild type cells. A substantial fraction of secreted coat proteins was loosely associated with the surface of the mutant cells, resembling the precoat posited to form early during normal sporulation. Transmission electron microscopy revealed that the precoat had little ultrastructural organization in the absence of cellulose. Thus, cellulose in the coat appears to be required for the organization of the pre-coat precursors as well as the stability, dormancy, and shape of the spore.

Percolation clustering: a novel approach to the clustering of gene expression patterns in Dictyostelium development.

We present a novel approach to the clustering of gene expression patterns based on the mutual connectivity of the patterns. Unlike certain widely used methods (e.g., self-organizing maps and K-means) which essentially force gene expression data into a fixed number of predetermined clustering structures, our approach aims to reveal the natural tendency of the data to cluster, in analogy to the physical phenomenon of percolation. The approach is probabilistic in nature, and as such accommodates the possibility that one gene participates in multiple clusters. The result is cast in terms of the connectivity of each gene to a certain number of (significant) clusters. A computationally efficient algorithm is developed to implement our approach. Performance of the method is illustrated by clustering both constructed data and gene expression data obtained from Dictyostelium development.

The membrane glycoprotein gp150 is encoded by the lagC gene and mediates cell-cell adhesion by heterophilic binding during Dictyostelium development.

gp150 is a membrane glycoprotein which has been implicated in cell-cell adhesion in the postaggregation stages of Dictyostelium development. An analysis of its tryptic peptides by mass spectrometry has identified gp150 as the product of the lagC gene, which was previously shown to play a role in morphogenesis and cell-type specification. Antibodies raised against the GST-LagC fusion protein specifically recognized gp150 in wild-type cells and showed that it is missing in lagC-null cells. Immunolocalization studies have confirmed its enrichment in cell-cell contact regions. In mutant cells that lack the aggregation stage-specific cell adhesion molecule gp80, gp150 is expressed precociously. Moreover, these cells acquire EDTA-resistant cell-cell binding during aggregation, suggesting a role for gp150 in this process. Cells in which the genes encoding gp80 and gp150 are both inactivated do not acquire EDTA-resistant cell adhesion during aggregation. Strains transformed with an actin 15::lagC construct express gp150 precociously, but do not show EDTA-resistant adhesion during early development. However, vegetative cells expressing gp150 can be recruited into aggregates of 16-h lagC-null cells. These results, together with those obtained with the cell-to-substratum binding assay, indicate that gp150 mediates cell-cell adhesion via heterophilic interactions with another component that accumulates during the aggregation stage.

The internal phosphodiesterase RegA is essential for the suppression of lateral pseudopods during Dictyostelium chemotaxis.

Dictyostelium strains in which the gene encoding the cytoplasmic cAMP phosphodiesterase RegA is inactivated form small aggregates. This defect was corrected by introducing copies of the wild-type regA gene, indicating that the defect was solely the consequence of the loss of the phosphodiesterase. Using a computer-assisted motion analysis system, regA(-) mutant cells were found to show little sense of direction during aggregation. When labeled wild-type cells were followed in a field of aggregating regA(-) cells, they also failed to move in an orderly direction, indicating that signaling was impaired in mutant cell cultures. However, when labeled regA(-) cells were followed in a field of aggregating wild-type cells, they again failed to move in an orderly manner, primarily in the deduced fronts of waves, indicating that the chemotactic response was also impaired. Since wild-type cells must assess both the increasing spatial gradient and the increasing temporal gradient of cAMP in the front of a natural wave, the behavior of regA(-) cells was motion analyzed first in simulated temporal waves in the absence of spatial gradients and then was analyzed in spatial gradients in the absence of temporal waves. Our results demonstrate that RegA is involved neither in assessing the direction of a spatial gradient of cAMP nor in distinguishing between increasing and decreasing temporal gradients of cAMP. However, RegA is essential for specifically suppressing lateral pseudopod formation during the response to an increasing temporal gradient of cAMP, a necessary component of natural chemotaxis. We discuss the possibility that RegA functions in a network that regulates myosin phosphorylation by controlling internal cAMP levels, and, in support of that hypothesis, we demonstrate that myosin II does not localize in a normal manner to the cortex of regA(-) cells in an increasing temporal gradient of cAMP.

The cellulose synthase gene of Dictyostelium.

Cellulose is a major component of the extracellular matrices formed during development of the social amoeba, Dictyostelium discoideum. We isolated insertional mutants that failed to accumulate cellulose and had no cellulose synthase activity at any stage of development. Development proceeded normally in the null mutants up to the beginning of stalk formation, at which point the culminating structures collapsed onto themselves, then proceeded to attempt culmination again. No spores or stalk cells were ever made in the mutants, with all cells eventually lysing. The predicted product of the disrupted gene (dcsA) showed significant similarity to the catalytic subunit of cellulose synthases found in bacteria. Enzyme activity and normal development were recovered in strains transformed with a construct expressing the intact dcsA gene. Growing amoebae carrying the construct accumulated the protein product of dcsA, but did not make cellulose until they had developed for at least 10 hr. These studies show directly that the product of dcsA is necessary, but not sufficient, for synthesis of cellulose.

An adenylyl cyclase that functions during late development of Dictyostelium.

A variety of extracellular signals lead to the accumulation of cAMP which can act as a second message within cells by activating protein kinase A (PKA). Expression of many of the essential developmental genes in Dictyostelium discoideum are known to depend on PKA activity. Cells in which the receptor-coupled adenylyl cyclase gene, acaA, is genetically inactivated grow well but are unable to develop. Surprisingly, acaA(-) mutant cells can be rescued by developing them in mixtures with wild-type cells, suggesting that another adenylyl cyclase is present in developing cells that can provide the internal cAMP necessary to activate PKA. However, the only other known adenylyl cyclase gene in Dictyostelium, acgA, is only expressed during germination of spores and plays no role in the formation of fruiting bodies. By screening morphological mutants generated by Restriction Enzyme Mediated Integration (REMI) we discovered a novel adenylyl cyclase gene, acrA, that is expressed at low levels in growing cells and at more than 25-fold higher levels during development. Growth and development up to the slug stage are unaffected in acrA(-) mutant strains but the cells make almost no viable spores and produce unnaturally long stalks. Adenylyl cyclase activity increases during aggregation, plateaus during the slug stage and then increases considerably during terminal differentiation. The increase in activity following aggregation fails to occur in acrA(-) cells. As long as ACA is fully active, ACR is not required until culmination but then plays a critical role in sporulation and construction of the stalk.

Cell-sorting in aggregates of Dictyostelium discoideum.

When Dictyostelium cells are induced to develop between a coverslip and a layer of agarose, they aggregate normally into groups containing up to a thousand cells but are then constrained to form disks only a few cells thick that appear to be equivalent to the three-dimensional mounds formed on top of agarose. Such vertically restricted aggregates frequently develop into elongated motile structures, the flattened equivalent of three-dimensional slugs. The advantage of using this system is that the restricted z-dimension enables direct microscopic visualization of most of the cells in the developing structure. We have used time lapse digital fluorescence microscopy of Dictyostelium strains expressing green fluorescent protein (GFP) under the control of either prestalk or prespore specific promoters to follow cell sorting in these flattened mounds. We find that prestalk and prespore cells expressing GFP arise randomly in early aggregates and then rotate rapidly around the disk mixed with the other cell type. After a few hours, the cell types sort out by a process which involves striking changes in relative cell movement. Once sorted, the cell types move independently of each other showing very little heterotypic adhesion. When a group of prestalk cells reaches the edge of the disk, it moves out and is followed by the prespore cell mass. We suggest that sorting may result from cell type specific changes in adhesion and the consequent disruption of movement in the files of cells that are held together by end-to-end adhesion.

tip genes act in parallel pathways of early Dictyostelium development.

Analysis of Dictyostelium strains carrying null mutations in tipA showed a primary defect in cell sorting and the formation of tips on the developing mound. To study the process affected in tipA- mutants further, other mutants with a similar phenotype were isolated and characterized. These studies showed three new Dictyostelium genes: tipB, tipC, and tipD. All the tip mutants aggregate into larger than average mounds, which split up and form many lips on their surfaces. Furthermore, each mutant exhibits reduced or aberrant cell-sorting behavior, never makes migrating slugs, and has severely reduced fruiting body and spore production. The mRNA of each tip gene is present in vegetative cells and does not vary significantly with development. Prespore and prestalk gene expression is reduced or delayed in the tip mutants indicating cell type differentiation is dependent on the function of these genes. Developing mutant cells in chimeric mixtures with wildtype cells demonstrated that the defects in each tip mutant behave cell autonomously. The overexpression of TipA in a tipB- background and the overexpression of TipB in a tipA- background significantly improved the morphogenesis of these mutants. These were the only situations in which the expression of one tip gene could compensate for the lack of a different tip gene. Except for the tipA-/tipB- strain, double mutations in the tip genes have additive effects, causing a more severe mutant phenotype with defects earlier in development than single mutants. The tipA-/tipB- double mutant does not show additive effects and is very similar to the tipA-single mutant. Analysis of the effects of double mutations and overexpression indicates that members of this class of genes appear to act through parallel pathways of differentiation and tip formation in early Dictyostelium development. Furthermore, TipA and TipB appear to have some overlapping functions or are involved in the same pathway. The multitipped phenotype observed in all the mutants may be a general result of perturbing early developmental events such as cell type differentiation and cell type proportioning.

SDF-2 induction of terminal differentiation in Dictyostelium discoideum is mediated by the membrane-spanning sensor kinase DhkA.

SDF-2 is a peptide released by prestalk cells during culmination that stimulates prespore cells to encapsulate. Genetic evidence indicates that the response is dependent on the dhkA gene. This gene encodes a member of the histidine kinase family of genes that functions in two-component signal transduction pathways. The sequence of the N-terminal half of DhkA predicts two hydrophobic domains separated by a 310-amino-acid loop that could bind a ligand. By inserting MYC6 epitopes into DhkA, we were able to show that the loop is extracellular while the catalytic domain is cytoplasmic. Cells expressing the MYC epitope in the extracellular domain of DhkA were found to respond only if induced with 100-fold-higher levels of SDF-2 than required to induce dhkA+ cells; however, they could be induced to sporulate by addition of antibodies specific to the MYC epitope. To examine the enzymatic activity of DhkA, we purified the catalytic domain following expression in bacteria and observed incorporation of labelled phosphate from ATP consistent with histidine autophosphorylation. Site-directed mutagenesis of histidine1395 to glutamine in the catalytic domain blocked autophosphorylation. Furthermore, genetic analyses showed that histidine1395 and the relay aspartate2075 of DhkA are both critical to its function but that another histidine kinase, DhkB, can partially compensate for the lack of DhkA activity. Sporulation is drastically reduced in double mutants lacking both DhkA and DhkB. Suppressor studies indicate that the cyclic AMP (cAMP) phosphodiesterase RegA and the cAMP-dependent protein kinase PKA act downstream of DhkA.

A molecular network that produces spontaneous oscillations in excitable cells of Dictyostelium.

A network of interacting proteins has been found that can account for the spontaneous oscillations in adenylyl cyclase activity that are observed in homogenous populations of Dictyostelium cells 4 h after the initiation of development. Previous biochemical assays have shown that when extracellular adenosine 3',5'-cyclic monophosphate (cAMP) binds to the surface receptor CAR1, adenylyl cyclase and the MAP kinase ERK2 are transiently activated. A rise in the internal concentration of cAMP activates protein kinase A such that it inhibits ERK2 and leads to a loss-of-ligand binding by CAR1. ERK2 phosphorylates the cAMP phosphodiesterase REG A that reduces the internal concentration of cAMP. A secreted phosphodiesterase reduces external cAMP concentrations between pulses. Numerical solutions to a series of nonlinear differential equations describing these activities faithfully account for the observed periodic changes in cAMP. The activity of each of the components is necessary for the network to generate oscillatory behavior; however, the model is robust in that 25-fold changes in the kinetic constants linking the activities have only minor effects on the predicted frequency. Moreover, constant high levels of external cAMP lead to attenuation, whereas a brief pulse of cAMP can advance or delay the phase such that interacting cells become entrained.

Cell-cell signaling during Dictyostelium development.

Specific proteins and peptides, as well as cAMP, are used as intercellular signals in Dictyostelium. Our understanding of the signal transduction pathways activated by these signals has been expanded by inclusion of newly characterized proteins. cAMP-dependent protein kinase (PKA) and its associated phosphodiesterase, RegA, play multiple roles in these pathways.

Role of PKA in the timing of developmental events in Dictyostelium cells.

The cyclic AMP (cAMP)-dependent protein kinase, PKA, is dispensable for growth of Dictyostelium cells but plays a variety of crucial roles in development. The catalytic subunit of PKA is inhibited when associated with its regulatory subunit but is activated when cAMP binds to the regulatory subunit. Deletion of pkaR or overexpression of the gene encoding the catalytic subunit, pkaC, results in constitutive activity. Development is independent of cAMP in strains carrying these genetic alterations and proceeds rapidly to the formation of both spores and stalk cells. However, morphogenesis is aberrant in these mutants. In the wild type, PKA activity functions in a circuit that can spontaneously generate pulses of cAMP necessary for long-range aggregation. It is also essential for transcriptional activation of both prespore and prestalk genes during the slug stage. During culmination, PKA functions in both prespore and prestalk cells to regulate the relative timing of terminal differentiation. A positive feedback loop results in the rapid release of a signal peptide, SDF-2, when prestalk cells are exposed to low levels of SDF-2. The signal transduction pathway that mediates the response to SDF-2 in both prestalk and prespore cells involves the two-component system of DhkA and RegA. When the cAMP phosphodiesterase RegA is inhibited, cAMP accumulates and activates PKA, leading to vacuolation of stalk cells and encapsulation of spores. These studies indicate that multiple inputs regulate PKA activity to control the relative timing of differentiations in Dictyostelium.

A cAMP-phosphodiesterase controls PKA-dependent differentiation.

A cAMP-specific phosphodiesterase was found that is stimulated by binding to the regulatory subunit of cAMP-dependent protein kinase, PKA-R, from either Dictyostelium or mammals. The phosphodiesterase is encoded by the regA gene of Dictyostelium, which was recovered in a mutant screen for strains that sporulate in the absence of signals from prestalk cells. The sequence of RegA predicts that it will function as a member of a two-component system. Genetic analyses indicate that inhibition of the phosphodiesterase results in an increase in the activity of PKA, which acts at a check point for terminal differentiation. Conserved components known to affect memory, learning and differentiation in flies and vertebrates suggest that a similar circuitry functions in higher eukaryotes.

Signal transduction pathways leading to spore differentiation in Dictyostelium discoideum.

Cells that overexpress PKA as a consequence of carrying multiple copies of the gene encoding the catalytic subunit can be induced to sporulate when developing as single cells. A peptide phosphorylated by PKA, termed SDF-1, has recently been shown to stimulate this process (Anjard et al., 1997). Several genes have been implicated in a signal transduction pathway by which prestalk cells induce encapsulation of prespore cells during terminal differentiation including a prestalk-specific putative membrane protease (TagC) and a two-component system consisting of a receptor-histidine kinase (DhkA) and a response regulator with cAMP phosphodiesterase activity (RegA). To determine whether SDF-1 uses this pathway, strains carrying null mutations in the pertinent genes were transformed with a pkaC plasmid such that they can overexpress PKA. Since these mutant strains all sporulated efficiently when SDF-1 was added, it appears that other gene products mediate the response. However, we found that regA- mutant cells release a distinct factor, SDF-2, that rapidly induces encapsulation of test cells overexpressing pkaC. Since cells in which tagC is disrupted do not form SDF-2 and cells in which dhkA is disrupted do not respond to SDF-2, this peptide appears to use the two-component system that regulates PKA activity. SDF-2 is a small peptide released by prestalk cells in a manner dependent on TagC. It appears to act on prespore cells through the DhkA receptor to inhibit the cAMP phosphodiesterase of RegA, thereby activating PKA via cAMP. The process of induction by SDF-2 can be shown to be distinct from that by SDF-1. SDF-2 appears to stimulate prestalk cells to release additional SDF-2 by acting through a signal transduction pathway that also involves DhkA, RegA, and PKA. Based on these results we present a model for the signal transduction cascade regulating spore differentiation.

Two-component signal transduction systems in eukaryotic microorganisms.

Conserved signal transduction pathways that use phosphorelay from histidine kinases through an intermediate transfer protein (H2) to response regulators have been found in a variety of eukaryotic microorganisms. Several of these pathways are linked to mitogen-activated protein kinase cascades. These networks control different physiological responses including osmoregulation, cAMP levels and cellular morphogenesis.