ABSTRACT
rZIP is an approx. 32 kDa, multi-domain protein of Dictyostelium discoideum whose structural motifs include a RING (zinc-binding) domain, a leucine zipper, a glutamine repeat, an SH3-binding region and a consensus phosphorylation site for MAP kinase. In vitro, rZIP forms homodimers and interacts specifically with the SH3 domain(s) of the Nck adaptor protein. rZIP is expressed maximally during cell differentiation at approximately equivalent levels in all cells. Disruption of the rZIP gene rzpA results in altered cellular aggregation, impaired slug migration, and aberrant patterning of prespore and prestalk cells, the major progenitor classes. In rzpA− strains, presporespecific genes are overexpressed and prestalk expression zones are reduced. Conversely, constitutive overexpression of rzpA markedly decreases prespore-specific gene expression and significantly increases the expression of prestalk-specific genes. Further, induced transdifferentiation of prespore cells into prestalk cells is inhibited in rzpA− slugs. In light of these patterning defects, we suggest that the RING/zipper protein rZIP plays an important role in early cell fate decisions in Dictyostelium, acting as a positive regulator of prestalk differentiation and an inhibitor of prespore differentiation.
INTRODUCTION
In many developmental systems, cell fate is determined by cellsurface receptor mediated regulation of downstream pathways. The relative activity of various protein kinases is often central to this regulation, but it is increasingly evident that communication among components within a pathway is often dependent upon additional associations. Many specific homologous and heterologous interactions involve a variety of motifs that are common to proteins of diverse function. In some instances, individual proteins possess more than one interaction domain, allowing multimeric associations among three or more components. In addition, protein-protein binding is often regulated by site-specific phosphorylation. Protein-interaction motifs that are required for signal control of differentiation include SH3 domains, SH2 domains (Pawson and Schlessinger, 1993), leucine zippers (Alber, 1992), and PH domains (Bork, 1996). Other modular sequences, such as the zinc-binding RING (C3HC4) domain (Saurin et al., 1996) and the glutamine repeat (Jennings, 1995), may be similarly essential. We have now identified rZIP, a new protein in Dictyostelium discoideum, that contains several of these structural motifs and regulates cell determination during development, possibly through a receptor mediated intracellular signalling pathway.
Dictyostelium has a well characterized life cycle in which growth and development are effectively independent (Firtel, 1995; Loomis, 1982). Under appropriate conditions, cells grow and divide continuously. As nutrients become scarce, a developmental process is induced that is completed within approx. 24 hours. Individual cells secrete cAMP and recruit neighboring cells by stimulation of G protein-linked, cell-surface cAMP receptors that relay this signal and direct chemotactic movement. Many subsequent aspects of differentiation and pattern formation throughout development are also regulated by these receptors that are linked to distinct pathways of intracelluar signalling (Ginsburg et al., 1995; Kimmel and Firtel, 1991). Cells converge at the centers of cAMP synthesis, form loose aggregation mounds, and differentiate into prespore and prestalk cells. As the aggregates become more compact, extracellular cAMP levels rise and the differentiated cell types sort from one another. An extended tip of prestalk cells forms and the mound elongates into a pseudoplasmodium or slug. The prestalk cells constitute the anterior (approx. 20%) of the slug and the prespore cells compose the posterior (approx. 80%). In addition, a small population of prestalk cells, the anterior like cells (ALCs), are dispersed through the prespore region. Upon completion of morphogenesis the slug forms a mature fruiting body, consisting of a stalk of vacuolated cells that supports a spherical mass of single-celled spores. The non-terminally differentiated prestalk and prespore cells, respectively, give rise to the stalk and spore cells. The basal disc at the foot of the stalk and the upper and lower cups, groups of cells which cradle the spore head, derive from the ALCs.
Here we describe rZIP, a modular protein that plays a role in regulating differentiation and patterning during Dictyostelium development. rZIP forms homo- and hetero-dimers in vitro and represses prespore but activates prestalk differentiation in vivo. The multiple domains of rZIP include a RING motif, a leucine zipper, a glutamine repeat, an SH3 binding region and a consensus phosphorylation site for MAP kinase. We suggest that these domains contribute to the complex signalling pathways during Dictyostelium development that originates with the surface receptors for cAMP and regulates cell determination. rZIP further provides an excellent system to examine the interdependence of these protein motifs in pattern formation.
MATERIALS AND METHODS
Growth and development
Dictyostelium were grown axenically with 100 μg/ml thymidine and 20 μg/ml G418 as required. Cells were plated for development on nitrocellulose filters at 3×106 cells/cm2 and on water-agar plates at 5.1×105 or 1.3×105 cells/cm2. For transdifferentiation, slugs were allowed to develop on agar for 16-18 hours and cut using aluminum foil.
Construction of rzpA− and rzpAOE strains and of the rzpA promoter/lacZ fusion
A BglII site was introduced into the leucine zipper region of rzpA and the THY1 gene was inserted into this site. This entire fragment was transformed into the thy1− strain JH10 (Saxe III et al., 1993). rzpA− transformants were identified by PCR and Southern analyses. For rzpA over-expression, a full-length rzpA clone fused behind the actin 15 promoter was transformed into wild-type Dictyostelium.
To isolate the promoter for rzpA, we subcloned the 2 kb EcoRI/BamHI fragment of the rzpA 5′ region from a genomic library. The BamHI site is 6 bp upstream of the translation start site and approx. 225 bp 3′ to the transcriptional start site. The EcoRI site lies at least 800 bp upstream of the translational stop codon of the adjacent upstream open reading frame (Ginsburg and Kimmel, unpublished). This EcoRI/BamHI fragment was fused to lacZ on a neomycin selectable vector.
β-galactosidase staining
Organisms on nitrocellulose were stained as described by Richardson et al. (1994). Vegetatively growing cells were allowed to adhere to a glass microscope slide for 30 minutes in a drop of axenic medium. The medium was then removed and the cells stained as before.
rZIP protein and antibody preparation and SH3 domain interaction
rZIP was expressed with a six-histidine amino-terminal tag using QIAexpressionist and the fusion protein purified by Ni2+ affinity chromotography (Qiagen). Rabbits immunized with the purified rZIP fusion produced polyclonal antibodies against rZIP. GST fusions with SH3 domains of Fyn, α-spectrin, phospholipase Cγ1, Eps8 and the three SH3 domains of Nck have been described previously (Rivero et al., 1994). Interaction studies between rZIP and these SH3 domains were conducted as described by Keane et al. (1995).
In vitro protein interaction
The leucine zipper/RING domain (see Fig. 1A) portion of rzpA was amplified by PCR and subcloned into pAR(ΔRI) (Blanar and Rutter, 1992). This construct directed expression of the leucine zipper/RING domain fragment fused to a FLAG recognition sequence (IBI) and a heart muscle kinase (HMK) phosphorylation site. The expressed peptide was purified using the FLAG system. Labelling of the peptide and binding studies were carried out as previously described (Blanar and Rutter, 1992; Li et al., 1992).
RNA analyses
Total RNA was isolated, size separated and hybridized to blots as described by Kimmel (1987).
RESULTS
rZIP is a modular protein
rzpA was isolated as a partial length cDNA (SC1) that is a member of a family of Dictyostelium genes containing the trinucleotide repeat (AAC)n (Kimmel and Firtel, 1985). We have now isolated full length cDNA and genomic DNA fragments of rzpA and determined their entire sequences. rzpA is a singlecopy gene in Dictyostelium (Kimmel and Firtel, 1985) and its cDNA and genomic sequences (GenBank accession number U67917) are colinear, indicating a lack of introns. The encoded protein, rZIP, is 271 amino acids (approx. 32 kDa) and contains several notable motifs, including a glutamine repeat, a RING (C3HC4) domain and a leucine zipper (Fig. 1A).
A perfect in-frame (CAA)14 within the translated region encodes the glutamine repeat Q14. Degenerate variations of (CAA)n give rise to runs of asparagines [(AAT)4] and threonines [(ACA/T)11]. Glutamine repeats are common to many developmentally regulated genes in Dictyostelium (Kimmel and Firtel, 1985), and other phyla (Bierkamp and CamposOrtega, 1993; Duboule et al., 1987; Wharton et al., 1985).
The RING (C3HC4) domain, located near the carboxyl terminus of rZIP, is a well conserved zinc-binding motif defined by the consensus sequence:
CX2CX9-39CX1-3HX2-3CX2CX4-48CX2C (see Fig. 1B). RING domains are essential for the activity of various developmentally regulated genes and are suggested to mediate protein-protein interaction (Saurin et al., 1996).
Leucine zippers mediate protein dimerization (Alber, 1992). A leucine zipper motif, defined by four leucyl residues repeated every seventh amino acid, is evident near the central region of rZIP. Helical wheel analysis indicates that the intervening sequences can be modelled into an amphipathic helix with an incremental angle of 103° yielding an exact overlay of the four leucines.
When the entire rZIP protein sequence was compared by BLAST analyses (Altschul et al., 1990) of various databases, most identities were aligned due to shared glutamine repeats or RING domains. The RING domains of c-Cbl, encoded by the mouse proto-oncogene c-cbl (Langdon et al., 1989) and the related human protein Cbl-b (Keane et al., 1995) have the greatest similarity to that of rZIP (Fig. 1B). These RINGs are approx. 40% identical (approx. 50% similar) in amino acid sequence with that of rZIP and share identical spacings of the critical C3HC4 sequences. Although rZIP differs greatly from the Cbl family outside of the RING motif, all three proteins have leucine zippers and proline-rich regions. In rZIP, the latter region forms part of the core for the consensus MAP kinase phosphorylation site PITP (see Fig. 1A).
rZIP will form homodimers in vitro
To analyze potential homologous interactions between Rzip molecules, we used an in vitro protein blot assay. The rzpA sequence was truncated just upstream of the encoded leucine zipper and downstream of the RING domain (see Fig. 1A) and fused in frame at its new 5′ terminus to sequences encoding a FLAG-tag and a phosphorylation site for heart muscle kinase (HMK; Blanar and Rutter, 1992). This HMK-zipper/RING fusion was expressed in E. coli, purified by affinity to FLAG antisera, and radiolabelled in vitro with HMK and [γ32P]ATP. In addition, full length rZIP was expressed in E. coli as an amino-terminal fusion with a six-histidine tag, purified by Ni2+ affinity chromatography, subjected to SDS gel electrophoresis and blotted to membrane filters. The bound protein was refolded in situ and incubated with the radiolabelled HMK-zipper/RING fragment. A specific complex was observed with full-length rZIP (Fig. 2A). Binding to the size marker proteins (see Fig. 2A) or to whole cell protein extracts of E. coli or Dictyostelium was not detected (data not shown). Truncated forms of HMK-rZIP containing just the glutamine repeat, leucine zipper or RING domain were unable to form specific associations with full-length rZIP (data not shown).
rZIP binds specific SH3 domains in vitro
The Src homology 3 (SH3) domain is a approx. 50 amino acid sequence that interacts with specific proline-rich sequences (Cichetti et al., 1992; Mayer and Eck, 1995). Protein complexes stablized by these interactions are essential for many aspects of intracellular signalling (Pawson and Schlessinger, 1993). Various members of the SH3 family bind distinct proline-rich sequences preferentially. Cbl-b and c-Cbl will bind specific SH3 domains in vitro, presumably through interaction of their proline-rich regions (Keane et al., 1995; Rivero et al., 1994). We now demonstrate that rZIP is similarly capable of forming specific complexes with a subset of SH3 domains. In particular we tested rZIP binding to a variety of expressed SH3 domains from the tyrosine kinase Fyn, phospholipase Cγ 1 (PLC), cytoskeletal protein spectrin (Spc), receptor tyrosine kinase substrate Eps8, and the adaptor protein Nck. Purified rZIP fusion protein was incubated in vitro with mass equivalent quantities of various glutathione-S-transferase (GST) fusions to the various SH3 domains that had been pre-bound to agarose-glutathione (GSH). After extensive washing, the proteins that were retained on the agarose-GSH were eluted in SDS buffer, separated by SDS-PAGE, and blotted for western analysis against polyclonal antisera to purified rZIP. rZIP displayed specific interaction with SH3 domains of only one protein, Nck (Fig. 2B). rZIP did not interact significantly with GST fusions to the SH3 domains of Fyn, PLCγ, Eps8, or Spc, or with the control construct carrying only the GST domain.
Temporal and spatial expression of rzpA
The temporal expression pattern of rzpA was determined by developmental northern blot analysis (Fig. 3A). Vegetatively growing cells (0 hours) expressed low levels of rzpA mRNA but levels increased substantially as the differentiated cells appeared during the mound stage (approx. 10 hours). Expression levels remained unchanged through the rest of development. Two different sized transcripts were detected at all developmental times examined, but since the gene does not contain any introns, they do not encode multiple forms of the rZIP protein.
To examine the cell-specific expression of rzpA we fused its full-length promoter to the reporter gene lacZ and monitored the expression and spatial patterning of the encoded β-galactoidase activity during Dictyostelium development. Northern analysis showed that a 2kb rzpA genomic fragment, including 225 bp of 5′-untranslated sequence and extending into the adjacent upstream open reading frame, directed fused lacZ expression in a pattern that is temporally similar to that of endogenous rzpA and consistent with the action of a full-length rzpA promoter (Fig. 3B). During growth (0 hr) when expression is very low, only approx. 5% of the cells had detectable β-galactosidase activity (Fig. 3C). When the strains were developed for various times, β-galactosidase staining was observed in all regions of the slug (16 hours) and the fruiting body (24 hours), indicating that rzpA is expressed in all celltypes throughout development (Fig. 3C).
rZIP is required for normal cellular motility during development
We created an rzpA− strain by homologous recombination within the endogenous gene of a wild-type organism. Three separate clonal isolates of rzpA− were confirmed by PCR and Southern and northern blot analyses (see Fig. 3A) and showed identical developmental phenotypes. Although the rzpA− strains were able to proceed through all the major stages of development, they displayed some prominent abnormalities. Their development was delayed by approx. 2 hours as compared with the wild-type strain, and during aggregation at a density of 1.3×105 cells/cm2 the rzpA− strains formed longer aggregation streams than did wild-type cells (Fig. 4A) and many fewer streaming centers (approx. 20% of wild-type; see Fig. 4A and Table 1). Rather than coalescing to a single center, individual rzpA− streams often fractured along their length and formed subsidiary aggregates of variable size (Fig. 4B). At a higher cell density (5.1×105 cells/cm2), where cell-cell communication was enhanced, the differences in aggregation patterns between the rzpA− and wild-type strains were less apparent (Table 1).
Slugs of the rzpA− strains were less elongate and much less motile than those of wild-type. When subjected to unidirectional light for 60 hours, wild-type slugs migrated an average distance of approximately 4 cm, whereas the rzpA− slugs were nearly immobile and often developed into fruiting bodies at the site of aggregation (Fig. 5).
Cell-type specific gene expression is perturbed in rzpA− strains
Developmental nothern blots of RNA from wild-type and rzpA− strains were hybridized with probes to prestalk and prespore cell-specific genes (Fig. 6). Expression of the prestalk-specific ecmA and ecmB mRNAs was slightly reduced in rzpA− strains relative to that of wild type, while expression of prespore-specific cotB, pspA and pspB mRNAs (data not shown for pspB) was both precocious and significantly higher in rzpA− cells.
The spatial patterns of cell-type specific gene expression in the rzpA− strains were examined by use of the lacZ reporter gene fused to cell-type specific promoters. In wild-type slugs carrying prestalk-specific ecmA/lacZ, β-galactosidase activity was concentrated in the anterior, prestalk zone, with the weaker activity in anterior like cells (ALCs) scattered through the posterior, prespore region (Fig. 7A-1). In contrast, prestalk ecmA/lacZ expression in rzpA− slugs was restricted to a smaller anterior zone and nearly absent in the ALCs. The lower cups and basal discs which derive from ALCs did not express β-galactosidase in rzpA− fruiting bodies marked with ecmA/lacZ (Fig. 7A-2). Similarly, prestalk ecmB/lacZ expression in rzpA− cells was not detected in ALCs distributed among the prespore cells, as observed with wild-type, but was restricted to the very rear of the slug (the rearguard zone) and to the central core of the anterior tip (Fig. 7B-1), and was reduced in the lower cups of the fruiting bodies (Fig. 7B-2).
In wild-type slugs carrying prespore cotB/lacZ, β-galactosidase was expressed primarily throughout the prespore zone, but additional weak blue staining was observed in some cells of the prestalk zone (Fig. 7C). It has been previously shown that this anterior staining resulted from the transdifferentiation of lacZ-expressing prespore cells into prestalk cells and their migration to the slug anterior (Detterbeck et al.,1994). Although the rzpA− slugs carrying the cotB/lacZ construct also displayed very strong βgalactosidase activity in the prespore region, staining was absent from the anterior and the very posterior of the slug. There was near mutual exclusion of the prespore (cotB) and prestalk (ecmA) expression patterns in the rzpA− slugs, suggesting that prespore to prestalk cell transdifferentiation was disturbed. The reduction in ALCs in the rzpA− strain was consistent with their postulated role as intermediates between prespore cells and anterior prestalk cells in the transdifferentiation pathway (Abe et al., 1994; Sternfeld, 1992).
rzpA− prespore cells are inhibited in transdifferentiation
The perturbations in spatial patterning of cellspecific expression in the rzpA− strains suggested that their prespore to prestalk cell transdifferentiation was disrupted. To examine this more directly, we sliced wild-type and rzpA− slugs transversely, just posterior to the prestalk/prespore boundary, and allowed the resultant ‘halves’ of each slug to resume development (Fig. 8). As previously shown, when the wild-type slugs were sectioned, each separate half formed a new slug with cell-type proportions that were similar to the original (Fig. 8; Raper, 1940). Since the back halves of slugs consisted almost entirely of prespore cells, a substantial transdifferentiation of prespore cells to prestalk cells was required to maintain constant prespore/prestalk proportions. Invariably, the front halves of the wild-type reformed slugs had completed development within 5 hours, approx. 5 hours earlier than did the rear halves. For Fig. 8, the front-half of the wild-type slug was removed from the microscope field.
The front halves of the rzpA− slugs reformed slugs in a manner similar to the front halves of wild-type but, not surprisingly, these resulting rzpA− slugs were unable to migrate (Fig. 8; see also Fig. 5). Although mutant structures were delayed by approx. 1 hour as compared with the front-halves of wild-type slugs, they still completed development approx. 4 hours earlier than did the rear-halves of the wild-type strain. In contrast, the rear-halves of rzpA− slugs remain arrested as mounds through the duration of the experiment (36 hours). If the rear halves of the rzpA− organisms were kept in the dark, they sometimes elongated into slugs. However, rear slug formation of rzpA− was delayed until 10 hours after cutting, and never progressed to the fruiting body stage. These results strongly suggest that rzpA− prespore cells are impaired in their ability to transdifferentiate from the prespore to prestalk state.
Overexpression of rZIP causes cell-specific effects of gene expression
The phenotypic effects of overexpression of rZIP in wild-type cells were investigated using the constitutive act15 promoter (Knecht et al., 1986). Northern analyses showed that the resulting act15/rzpA (rzpAOE) transformants overexpressed rzpA mRNA >100-fold (Fig. 9A). Western analyses similarly confirmed overexpression (>30-fold) of the rZIP protein (data not shown).
The rzpAOE strains were developed for 60 hours in unidirectional light as an assay for slug migration (Fig. 5). These slugs exhibited haphazard migration and made frequent sharp turns away from the light source, suggestive of impaired phototaxis. This was distinct from the straight movement of wildtype slugs toward unidirectional side light (see Fig. 5). rzpAOE strains remained as slugs for longer times than did the wildtype controls under all light conditions examined (darkness; unidirectional side light; overhead light), often not culminating until 10 hours after wild-type in overhead light.
Northern analyses showed dramatic effects of rzpA over-expression on cell-specific gene expression. The presporespecific markers cotB and pspA (data not shown) were substantially underexpressed in rzpAOE strains compared to controls, whereas the prestalk markers ecmA and ecmB were overexpressed (Fig. 9B). In wild-type, ecmB was not expressed until the tight aggregate stage (approx. 12 hours), but in the rzpAOE strain, expression was even detected during vegetative growth.
Prestalk-specific ecmA/lacZ and ecmB/lacZ, and presporespecific pspA/lacZ constructs were used to examine the regulation of cell-type patterning in the rzpAOE strain. The anterior region of expression of ecmA/lacZ in rzpAOE strains was expanded compared with the wild-type (Fig. 10A; see Fig. 7A-1). rzpAOE strains carrying the prestalk ecmB/lacZ had more intense β-galactosidase activity than did wild-type cells but it was concentrated toward the posterior of the slug (Fig. 10B), consistent with an expanded anterior prestalk A zone in this strain (see Figs 10A, 7A-1, and 7B-1). In contrast, very little β-galactosidase activity was detected with prespore pspA/lacZ in an rzpAOE strain as compared with wild-type slugs (Fig. 10C), reflecting reduced expression of the endogenous pspA mRNA in rzpAOE cells (see Fig. 9B).
DISCUSSION
The RING/zipper protein rZIP plays an important role in determining cell fate decisions during Dictyostelium development. This may involve regulatory pathways requiring complex interactions among various homologous and heterologous protein molecules through glutamine repeat, RING, zipper and SH3-binding motifs. rZIP function could be additionally modulated by protein kinases whose activities depend upon intracellular controls that originate with surface receptor activation.
The glutamine repeat is common to many developmentally regulated genes of most eukaryotes. In Dictyostelium, these genes may encode transmembrane (e.g. the cAMP receptor subtype 4; Louis et al., 1994), cytosolic (e.g. the PKA catalytic subunit; Mann and Firtel, 1991), or nuclear (e.g. the transcription factor GBF; Schnitzler et al., 1994) proteins. Although the precise molecular function of the glutamine repeat is not defined, it can serve as an activation domain for several transcription factors in yeast (Gerber et al., 1994). Glutamine repeats promote protein-protein dimerization in vitro (Stott et al., 1995), and inappropriate interactions due to an expanded glutamine repeat in the huntingtin protein are suggested to underly the neurodegenerative Huntington’s disease (Burke et al., 1996; Li et al., 1995).
In addition to the glutamine repeat, the leucine zipper and RING domain may also regulate macromolecular interactions of rZIP. While two structurally similar leucine zippers are necessary and often sufficient for dimerization, some zipper proteins may require an additional domain (e.g. HLH; helixloop-helix) for functional associations. Initial studies of rZIP suggest that both the leucine zipper and RING domain may contribute to its homodimer formation (Fig. 2A).
More than 80 proteins with extremely varied functions possess a zinc-binding RING (C3HC4) domain (Saurin et al., 1996). Unlike classical zinc fingers which are definitive DNA-binding moities, RING family members may be nuclear, cytosolic, or integral membrane proteins. Consistent with the length and sequence heterogeneity in the consensus motif defining the family (Fig. 1B), various members have different three-dimensional organizations (Borden et al., 1995) which may reflect their apparent functional diversity. It has been suggested that RING domains promote protein-protein associations, and perhaps different subfamilies have distinct structures and interactions.
The RING domains that are most similar to rZIP are on murine c-Cbl (Langdon et al., 1989), human Cbl-b (Keane et al., 1995), and SLI-1 (Yoon et al., 1995), a related C. elegans protein (Fig. 1B). SLI-1 has been implicated in a receptor tyrosine kinase mediated signalling pathway that regulates vulval development (Yoon et al., 1995) and the Cbl proteins, which like rZIP have a leucine zipper and SH3-binding domain (Keane et al., 1995; Rivero et al., 1994), are implicated in EGF receptor signalling (Galisteo et al., 1995). The v-cbl oncogene encodes a truncated version of c-Cbl that contains only the amino-terminal 40% of the wild-type cellular protein (Blake and Langdon, 1992). The deleted sequences include the RING domain, the leucine zipper, and the SH3 binding site(s). Correlated with this deletion is a subcellular redistribution of Cbl protein sequences from the cytoplasm to the nucleus and an acquisition to induce acute transformation (Langdon et al., 1992). By analogy we suggest that the RING, zipper and SH3-binding sites of rZIP are similarly important for control of celltype determination through a receptor-regulated intracellular signalling cascade during Dictyostelium development.
Expression of rzpA in Dictyostelium may be limited to a subset of vegetatively growing cells, but later in development all of the major differentiated cell types appear to express rzpA (see Fig. 3). Consistent with this temporal expression pattern, rzpA− strains exhibit aberrant aggregation very early in development and altered patterning of prespore and prestalk cells during later stages. The prespore genes pspA and cotB are induced precociously and to abnormally high levels in rzpA− lines and are underexpressed in rzpAOE lines which constitutively overexpress rzpA. In contrast, prestalk-specific gene expression in rzpA− lines is reduced relative to that of wildtype, while rzpAOE strains overexpress the prestalk markers ecmA and ecmB. Further, ecmB expression is extremely precocious in rzpAOE lines. Thus rzpA appears to be involved in a negative regulatory pathway for prespore differentiation but a stimulatory circuit for prestalk differentiation. These defective regulatory circuits for cell differentiation in rzpA− strains severely compromise the ability of prespore cells to transdifferentiate into prestalk cells. Perhaps rZIP plays a significant role in repression of prespore expression during this transdifferentiation.
As wild-type slugs migrate, they slough a population of prestalk cells into the slime trail. To maintain constant proportioning of their differentiated cell-types throughout development, there is a required redifferentiation of prespore cells into anterior-like cells (ALCs) and finally into anterior, prestalk cells (Abe et al., 1994; Sternfeld, 1992). Thus after prolonged migration, wild-type slugs that specifically express stable β-galactosidase in prespore cells exhibit staining in their anterior regions (Detterbeck et al., 1994; Harwood et al., 1991). In contrast, comparably marked rzpA− slugs have a sharp delineation of β-galactosidase staining at their prestalk/prespore boundaries (Fig. 7) and rzpA− strains which carry the prestalk-specific ecmA/lacZ or ecmB/lacZ show only very limited β-galactosidase staining in ALCs and ALC-derived structures as compared with an equivalent wild-type strain (Fig. 7A,B). However, the inability of rzpA− strains to undergo appropriate transdifferentiation is not solely due to poor slug migration. Unlike the wild-type strains, transected rzpA− slugs will not transdifferentiate a sufficient population of prespore cells into prestalk cells to permit continued development of posterior segments (Fig. 8). A dependent link may, therefore, exist among the events of transdifferentiation, cell movement within the slug, and migration of the slug. Thereby impaired transdifferentiation may abrogate slug migration.
The most posterior regions of cotB/lacZ marked rzpA− slugs stain poorly for β-galactosidase activity (Fig. 7C). The absence of rZIP may lead to the disruption of a morphogenic gradient within the prespore region of the slug, perhaps through alteration in production or perception of the cAMP signal. Further, the impaired phototaxis of rzpAOE slugs (Fig. 5) is very similar to that of slugs which lack active cAMP-dependent protein kinase (PKA) in prestalk cells (Bonner and Williams, 1994). Other data also support a defective cAMP signalling pathway in rzpA− cells. During aggregation at a low cell density (1.3×105 cells/cm2) they form fewer aggregation centers and much longer aggregation streams than do wild-type cells (Fig. 4, Table 1). Similar abnormal patterns of aggregation have been observed in mutants with defined defects in cAMP signal-relay (Gerisch, 1971; Riedel et al., 1973) and can be mimicked by specifically inhibiting cAMP signalling in wild-type cells by exogenous drug treatment (Newell and Ross, 1982). Furthermore, when rzpA− cells are plated at a higher cell density (5.1×105 cells/cm2) to enhance endogenous cAMP signalling, a more normal aggregation phenotype is observed.
The regulation of prestalk and prespore differentiation in Dictyostelium involves complex interactions among a variety of activation pathways and lateral inhibitions (Loomis 1993). Although the position of rZIP within these regulatory circuits is yet to be resolved, there are several characterized genes with which rZIP is likely to interact, either directly or indirectly. Strains which lack a functional cAMP receptor subtype 4 (CAR4) express higher prespore-specific and lower prestalkspecific mRNA levels than do wild-type strains (Louis et al., 1994). Thus, like rZIP, CAR4 inhibits prespore but stimulates prestalk differentiation. In addition, slugs of both car4− and rzpA− strains have reduced ALC expression and poor motility. These phenotypic similarities strongly suggest an involvement of rZIP in an intracellular signalling pathway that originates with cAMP activation of CAR4 at the cell surface. However, since rzpA transcripts are detectable in vegetative cells (see Fig. 3A), whereas car4 mRNA is not expressed prior to tight mound formation (Louis et al., 1994), it is not surprising that the rzpA− and car4− strains are not precise phenocopies of one another. By analyzing gene expression and cellular patterning in strains in which both genes are either disrupted and/or misexpressed, we will be able to define their epistatic relationship more clearly.
At the intracellular level, essential functions for three protein kinases [glycogen synthase kinase 3 (GSK3), cAMP-dependent protein kinase (PKA), and MAP kinase 2 (ERK2)] have been described for prespore/prestalk differentiation of Dictyostelium. Strains that lack GSK3 overexpress prestalk ecmB and underexpress prespore cell-specific markers (Harwood et al., 1995). Preliminary data also suggest an upstream involvement of CAR4 in GSK3 regulation. Both CAR4 and rZIP could be antagonistic to GSK3 in prestalk/prespore determination pathways. A model outlining these possible interactions is shown in Figure 11.
PKA is required for both prestalk and prespore expression (Mann and Firtel, 1993). Specific inhibition of PKA in prestalk cells blocks continued prestalk differentiation (Zhukovskaya et al., 1996) and terminal differentiation of the stalk (Harwood et al., 1992). Similarly, prespore cells which express a dominantnegative regulatory subunit of PKA become arrested and do not progress towards spore cell differentiation (Hopper et al., 1993). Since expression of ERK2 in prespore cells is required for continued prespore-specific gene expression (Gaskins et al., 1996), and since rZIP will downregulate this expression pattern, these proteins may also participate in a linked regulatory pathway. Phosphorylation by ERK2 of the consensus MAP kinase site within the proline-rich region of rZIP may alter the interaction of rZIP with an endogenous SH3-domain partner to potentiate prespore differentiation. Currently the extent of the SH3 population in Dictyostelium is unclear. The in vitro binding data of rZIP in conjunction with the more extensive studies of Cbl suggest that rZIP could function in a multimeric complex involving an Nck-like SH3-SH2 adaptor protein. Alternatively rZIP may be a novel adaptor. Its multiple protein interacting domains could underlie a new intracellular signalling mechanism regulated by protein phosphorylation.
The overall regulatory modes for prestalk and prespore determination in Dictyostelium are mirrored in many other developmental systems. GSK activity is critical for dorsal/ventral patterning in Xenopus (Pierce and Kimelman, 1995; Xi et al., 1995) and for establishing segment polarity in Drosophila (Perrimon, 1994). PKA and MAP kinases play essential roles in general aspects of cell growth and development. Central to the activation or inhibition of these pathways is a transmembrane signalling cascade. A similar linkage between the cAMP receptors and rZIP of Dictyostelium may recall or predict related interactions in other organisms. It is now essential to define the genetic heirarchies and physical associations by which rZIP regulates cell fate.
ACKNOWLEDGEMENTS
We thank Drs J. Williams and R. Firtel for plasmid constructs. We are grateful to Drs R. Insall, M. Blanar, J. Louis, K. Rogers, D. Richardson, B. Oliver, M. Krause and D. Ratner for their comments and suggestions.