The Dictyostelium bZIP transcription factor DimB regulates prestalk-specific gene expression

Primary patterning of the Dictyostelium slug creates an anterior prestalk region and a posterior prespore region but, in addition to the coherent mass of prestalk cells that comprises the front one-fifth of the slug, there are Anterior-Like Cells (ALC) scattered throughout the prespore region. Distinct prestalk and ALC subtypes have been identified using ecmA, a gene that encodes an extracellular matrix protein.

ecmA is more strongly expressed in cells in the front of the prestalk region than in the back and cells in these two locations use spatially separated regions of its promoter (Williams et al., 1989; Early et al., 1993). PstA cells occupy the front half of the prestalk region and employ cap-site proximal promoter elements, while pstO cells occupy the rear half and use cap-site distal promoter elements. Although they were initially identified using the bi-partite promoter of the ecmA gene, many other genes are expressed only in one or other region (Maeda et al.,2003). This indicates that the two subtypes are of wide developmental significance and the fact that they differ in their movement patterns within the slug and at culmination supports this notion(Jermyn and Williams, 1991; Abe et al., 1994).

The chlorinated hexaphenone DIF-1 rapidly induces ecmAtranscription (Williams et al.,1987) and dmtA-, a biosynthetic mutant which makes little or no DIF-1, shows a major defect in prestalk cell differentiation(Thompson and Kay, 2000b);pstO-specific gene expression is greatly reduced but pstA differentiation is unaffected. Thus, DIF-1 appears to be essential for efficient pstO cell differentiation but not for pstA cell differentiation. A genetic approach has identified a gene, dimA, that is required for multiple aspects of DIF-1 signalling (Thompson et al.,2004) but its interface with the DIF-1 signalling pathway is unknown.

The protein encoded by dimA is a bZIP transcription factor. bZIP proteins are found in all eukaryotes and are characterised by the presence of a basic, DNA-binding region and a closely apposed leucine zipper(Hai and Hartman, 2001; Jakoby et al., 2002; Poels and Broeck, 2004). The leucine zipper mediates the formation of homodimers and heterodimers with other bZIP proteins. We identify a novel bZIP protein that interacts with the ecmA promoter and show that it is a DIF-1 responsive regulator of prestalk specific gene expression.

Dictyostelium strain Ax2 was grown axenically and transformed by electroporation or, in the case of double transformants, using calcium phosphate precipitation (Nellen et al.,1984). Transformant pools were selected at 50 μg/ml G418 and stained for β-galactosidase activity or double-stained forβ-glucoronidase and β-galactosidase(Early et al., 1993).

Cells were harvested from growth, washed and plated at densities varying between 10⁴ and 10⁵ cells/cm² in stalk medium[10 mM KCl, 2 mM NaCl, 1 mM CaCl₂, 10 mM MES (pH 6.2)] and incubated at 22°C for 1 hour. The medium was changed to stalk salts,cerulenin at concentrations between 50 μM and 150 μM, and 5 mM cAMP, and was incubated for a further 6 hours. The medium was then removed and replaced with stalk medium, the same concentration of cerulenin(Kay, 1998) and concentrations of DIF-1 up to 100 nM. The plates were incubated at 22°C for 16 hours and RNA was extracted using an RNeasy kit (Qiagen) and analysed by RT-PCR using a`TITANIUM' One-Step RT-PCR kit (BD Biosciences). The ecmA primers are: forward, CCAATTAGCTGTCCAAAACC; reverse, GCAATCACCTTTACCTCCTG. They generate a 480 nucleotide fragment. IG7, a constitutively expressed mRNA, was used as control.

Nuclear extracts were prepared from slug stage cells(Kawata et al., 1996). For purification, nuclear extract derived from 3.5×10¹¹ slug cells was precipitated with 40% (w/v) ammonium sulfate, and subjected to heparin sepharose chromatography. It was then loaded onto a DNA affinity column bearing R1 or R2 oligonucleotides. These were synthesized as a duplicate tandem copy, multimerized by ligation and coupled to sepharose beads. Bound proteins were eluted with 0.4 M KCl. The eluted proteins were further purified through a second round of binding on the affinity column,concentrated and loaded onto an SDS gel. After staining with Colloidal Blue Staining Kit (Invitrogen), protein bands were excised from the gel, digested with trypsin and analysed by MALDI-TOF mass spectrometry.

The DimB-coding region was cloned in pET28a (Novagene), with the addition of a BamHI site at the N terminus and an XhoI site at the C terminus. It was expressed in E. coli strain BL21 CodonPlus RIL(Stratagene), as a C terminal HIS tagged fusion protein and purified using a`Talon' metal affinity column (Clontech). Gel retardation assays were performed, as described above, using approximately 1 μg of fusion protein per assay.

DimB-coding sequences extending from +1 to +1375 (numbered relative to the ATG) was cloned into a plasmid vector and a BamHI hygromycin resistance cassette was inserted at the unique BglII site, positioned at nucleotide 600 relative to the initiation codon. Transformants were isolated clonally and screened for gene disruption by PCR and by western blotting.

Polyclonal rabbit antisera were generated using the N terminal and C terminal 15 amino acids of DimB as immunogens. The peptides contained a non-coded cysteine residue, respectively at their C and N termini, and these were coupled to an affinity matrix in order to purify the antibody.

Ax2 cells were developed on non-nutrient water agar to form loose aggregates and DmtA null cells were developed to the tight aggregate stage. Aggregates were mechanically dissociated in KK2 buffer, by trituration through a syringe needle. After dissociation, cells were adjusted to a density of 2×10⁷ cells in KK2 and treated with 0.01% ethanol or with 100nM DIF-1 in 0.01% ethanol. Cells were fixed on ice in methanol for 10 minutes and stained with affinity purified, anti-DimB antibody. After secondary reaction with Alexa Fluor 488-conjugated goat anti-rabbit antibody(Molecular Probes), samples were analyzed by confocal microscopy.

Disaggregated cells, from the loose aggregate stage, were induced by shaking at 4×10⁶/ml in phosphate buffer containing 1 mM cAMP and 100 nM DIF-1 for 1.75 hours. They were then fixed with paraformaldehyde(final 1%) in PBS at 4°C for 2 hours. After stopping the reaction with glycine (final 125 mM) and washing, cells were suspended in TNTE buffer [50 mM Tris-Cl (pH 8.0) containing 150 mM NaCl, 1% Triton X100, 0.05% SDS, 2 mM EDTA]containing complete protein inhibitor cocktail (Roche Diagnostics, Germany)and sonicated to yield fragments of 0.3 to 3.0 kb. Immunoprecipitation was carried out from the supernatant in TNTE buffer containing 1% blocking reagent(Roche Diagnostics, Germany): using the anti-DimB antibody or an anti-CudA antibody, and Protein A-Sepharose CL-4B (Amersham Biosciences, Sweden) at 4°C, overnight.

After washing with TNTE buffer and then with 10 mM Tris-Cl (pH 8.0)containing 250 mM LiCl, 1% NP-40, 1% DOC and 1 mM EDTA, DNA was recovered by incubating with 50 mM Tris-Cl (pH 7.5) containing 1% SDS and 1 mM EDTA at 37°C for 15 minutes. After reversing the crosslinking by heating at 65°C overnight and treating with proteinase K (100 μg/ml) and RNase (10μg/ml) at 37°C for 1 hour, DNA was purified with a QIA Quick-Spin column (QIAGEN). PCR was performed from immunoprecipitated DNA or from total genomic DNA using the following primers: ecmA, AATTATAACCCCCATTCGC and AGAGTTTGATGATAACAAGAG (region -1257 to -876); and (as a control) the gpbA upstream region, TAAACAAACACACACCCAAC and AGGACTTACTAAAATTACAGG.

By combining known expression patterns of ecmA promoter deletion constructs (Early et al., 1993)with a limited characterisation of several new deletion constructs (M.F. and J.G.W., unpublished), we designed two oligonucleotides for use in affinity chromatography (Fig. 1).

A 5 to 3′ deletion construct of the ecmA promoter with a cap-site distal end point at -1041 [construct M(Early et al., 1993)] directs high level expression in pstA and pstO cells. Further deletion, to -895[construct N (Early et al.,1993)], reduces expression, equally in pstA and pstO cells. Hence,oligonucleotide R1 was designed, containing 32 nucleotides of sequence closely bordering the 5′ boundary of construct M. Deletion to -531 (construct O)generates the `standard' pstA marker and further deletion to -374 (construct P) greatly reduces expression (Early et al., 1995). Hence, oligonucleotide R2 was designed, containing 32 nucleotides of sequence closely bordering the 5′ boundary of construct O.

R1 was used to isolate interacting proteins, by employing its multimerised form in affinity chromatography with slug cell nuclear extracts. After elution the bound proteins were subjected to SDS gel-electrophoresis and the most abundant species identified by mass spectrometry(Fig. 2). Several are RNA-interacting proteins that bind non-specifically to the column. One is a novel bZIP protein that we term DimB (encoded by dimB). Another,lower molecular weight species on the gel, DimB′, contains a subset of the DimB peptides and is presumably a degradation product of DimB.

When the R2 region was used in affinity chromatography, multiple species were also purified. Again, mass spectrometry showed that several are RNA-binding proteins but one is DimB (Fig. 2). Another of the proteins (asterisked) binds to R2 but not to R1. It is a novel MYB domain protein, that we have named MybE (M.F. and J.G.W., unpublished).

DimB bears significant homology to DimA(Thompson et al., 2004) but is half its size; DimB has a predicted size of 68,447 Da, while DimA is predicted to be 140,259 Da (Fig. 3A). One region of strong homology between the two proteins encompasses the DNA binding/dimerisation domain and an alignment of the basic regions and leucine zippers of DimB, DimA and several other members of the bZIP family is presented in Fig. 3B.

DimB was expressed in E. coli as part of a fusion protein,DimB:HIS, in which oligo-histidine is linked to the C terminus. DimB:HIS was employed in band shift assays with either R1 or R2 as labelled probe. DimB:HIS binds to both the R1 and R2 probes, and unlabelled R1 and R2 oligonucleotides are potent inhibitors of the binding: both in self-competition and in cross-competition (Fig. 4; data not shown for R1). Just 10 pmoles of unlabelled R1 or R2 DNA strongly competes for binding of DimB:HIS to a labelled R1 or R2 probe. As negative controls, a discoidin 1γ promoter oligonucleotide, the TTG box(Vauti et al., 1990), and an ecmB promoter oligonucleotide, the ecmB activator(Ceccarelli et al., 2000), were used as competitors. Both are much weaker competitors than R1 or R2. Thus,recombinant DimB binds with high relative affinity to R1 and R2.

The positions of the DimB-binding sites in R1 and R2 were determined by a mutational scanning technique (Ceccarelli et al., 2000), in which four nucleotide blocks of sequence were replaced with the arbitrarily chosen sequence GCGC. The mutant forms were used as competitors in band-shift assays against the respective unmutated probe. In multiple experiments, using varying amounts of competitor, the M6 mutant form of R1 (data not shown) and the M7 mutant form of R2 consistently proved to be relatively ineffective as competitors (Fig. 5A).

Combining the above information with the sequences of several known bZIP binding sites allowed us to deduce a provisional binding consensus(Fig. 5B). The R1 scanning mutation analyses identified the invert complement of the sequence CATC and this forms part of the sequence CATCA. The R2 scanning analyses identified the related sequence CACA. Together with the cross-species comparison, this suggests that the consensus sequence necessary for binding is A-c/a-A-C-a/c-t/a-C-A (Fig. 5B).

Guided by this consensus, point mutations were inserted into R2. Point mutants R2pM2 and R2pM3, in which one or other of the cap-site distal C residues is changed to A, display reduced binding to DimB(Fig. 6A). This confirms the scanning mutation analysis, in showing that the sequence CACA is necessary for optimal binding of DimB. Point mutations that change all four of the C residues within the proposed consensus region to G, in oligonucleotide R2pM1,eliminate competition activity (Fig. 6A).

R2pM1, the oligonucleotide with the four point mutations that completely abrogate DimB binding, was cloned within a lacZ fusion construct to assay its function in vivo. This recipient construct, a newly created lacZ fusion termed S (Fig. 1, Fig. 6B), has its cap-site distal terminus at nucleotide -493. It differs significantly from published ecmA promoter fragment constructs at its cap-site proximal end (legend to Fig. 6B).

Fusion of R2 to S generates a construct, R2S, that is most strongly expressed in the pstA region but that is also expressed in cells in the pstO region and in large numbers of ALC (Fig. 6B). When this pattern is compared with S and R2pM1S, a striking difference is apparent. With the latter two constructs staining is almost entirely restricted to the pstA region, with very few stained cells in the pstO region and almost no stained ALC (Fig. 6B). Thus, the cap-site proximal (-510) DimB-binding site is not required for expression in pstA cells but is essential for efficient expression in pstO cells and ALC.

Polyclonal antibodies were raised against an N terminus-proximal and a C terminus-proximal peptide of DimB and purified by affinity chromatography on the respective peptide immunogen. The N terminus proximal-antibody recognises DimB and a higher molecular weight species (data not shown), and was used only to characterise potential null strains (see below). The C terminus-proximal antibody was used in western transfer to obtain a developmental time course of DimB accumulation (Fig. 7). The only species recognised by this antibody is a protein of 70 kDa, the approximate size of DimB. DimB is present at very low concentration at zero hours, rises in concentration to the tight mound stage and then remains at a relatively constant concentration throughout subsequent development.

In order to analyse DimB function genetically, a disruption construct, in which a hygromycin resistance cassette interrupts dimB, was transformed into cells. Resistant colonies were screened by PCR and clones with an apparent disruption of the dimB gene were examined by western transfer, using the N and C terminus-specific antibodies. In wild-type cells both antibodies recognise DimB but in the candidate null cells there is no signal at this position (Fig. 8A and data not shown for the N terminus-derived antibody). Thus, dimB is functionally inactivated in the disruptant strains (termed dimB-null or dimB- strains).

Early development and culmination of the null strains appear normal but at the slug stage several differences from control, random-integrant slugs become apparent. The null strains form elongated slugs(Fig. 8B, upper). When control cells are spotted on agar and exposed to a dim unidirectional source, the slugs that are formed move towards the light source(Fig. 8B, lower). DimB-slugs do not leave the point of origin. Examination of the point of origin suggests that this is not caused by a failure to enter the migratory slug stage. It would appear to reflect a defect in the intrinsic ability of the slug to move.

In order to analyse total prestalk cell differentiation, a dimB-strain and a random integrant strain were transformed with ecmAO:lacZ. After development to the standing slug stage, the slugs were stained forβ-galactosidase. A control and a mutant field, each containing three slugs, is shown in Fig. 9. These particular control and mutant slugs appear similar in the average relative sizes of their prestalk regions. This conclusion is supported by quantitative analysis of a larger number of slugs, but this revealed a higher level of heterogeneity in the prestalk: prespore ratio in the mutant slug population (explained in the legend to Fig. 9).

Having failed to detect a consistent difference in the relative size of the total prestalk population in the mutant, the prestalk subtypes were analysed. Cells were transformed with ecmA:lacZ [construct O(Early et al., 1993)],ecmO:lacZ and ecmB:lacZ. Each was a co-transformation with pspA:gluc, a construct in which the promoter of the pspA, prespore-specific, gene drives expression of β-glucoronidase. Staining was performed at the standing slug stage of development, using a red β-galactosidase chromogen and a blue β-glucoronidase chromogen (Early et al., 1993).

pspA:lacZ is, as expected, expressed in the rear of the slug and expression completely abuts the prestalk zone (Fig. 9). As in the control slugs, ecmA:lacZ is most highly expressed in the pstA region and ecmO:lacZ is expressed selectively in the pstO cells. The presence of a pstO zone is completely reproducible from slug to slug, and is of critical importance because it constitutes a clear difference from the dimA and dmtA null strains.

The ecmB:lacZ construct is expressed in a cone of pstAB cells within the slug tip and in the group of pstB cells, situated close to the prestalk-prespore boundary (Jermyn and Williams, 1991). Analysis of control and mutant ecmB:lacZ transformants at culmination shows the expected pattern, with strong staining in the stalk and the cups that cradle the spore head (data not shown).

The above results were obtained using cells developed under overhead light and analysed at the standing slug stage. In addition, the staining times were of 1-2 hours, by which time the enzymatic reactions for the ecmAO:lacZ transformants were approaching a plateau. Using ecmAO:lacZ marked slugs developed in the presence of a dim uni-directional light source, and with shorter (c. 5-10 minutes) staining times, an anteroposterior gradient of staining becomes apparent in the control(Fig. 10). These short staining times also show that staining in control slugs is stronger at the periphery of the prestalk region than in the core. The dimB- slugs display a very different pattern from the control slugs, their staining is uniform throughout the prestalk region.

DIF-1 responsiveness in the dimB- mutant was assayed using a protocol in which cells starved in the presence of cAMP were treated with various concentrations of DIF-1. Expression of the ecmA gene was monitored by RT-PCR and, as a quantitation control, parallel RT-PCR was performed for Ig7: a constitutively expressed mRNA. In order to minimise the effects of endogenously produced DIF-1, the assays were performed at low cell densities (10⁴ or 10⁵ cells/cm²) and the polyketide synthesis inhibitor cerulenin was included at various concentrations (Kay, 1998). Under all conditions tested, there was robust induction of ecmAexpression in the random integrant cells but no response in the dimB-cells (Fig. 11).

We also analysed cells that were rendered competent to respond to DIF-1 by allowing them to develop normally to the loose aggregate stage. After disaggregation, the cells were shaken in suspension with and without DIF-1 for 2 hours; ecmA expression was activated in the control cells but not in the dimB- cells (data not shown).

In order to determine whether DIF-1 directly regulates DimB loose aggregate, parental cells (data not shown) or tight aggregate stage cells derived from the DIF-1 deficient mutant dmtA-(Fig. 12) were exposed to DIF-1 and subjected to immunohistochemical staining. Similar results were obtained with both strains. There is a rise in nuclear staining, with a peak three minutes after DIF-1 addition.

In order to determine whether DIF-1 induces DimB to bind to the ecmA promoter in vivo, ChIP analysis was performed. In initial experiments, cells dissociated at the loose aggregate stage were incubated for several hours with cAMP, DIF-1 or cAMP and DIF-1. RT-PCR analysis was then used to monitor ecmA expression. As was observed previously(Berks and Kay, 1990), a combination of cAMP and DIF proved most effective and these conditions were employed for the ChIP assay.

Chromatin extracted from induced and control cells was precipitated with anti-DimB antibody or with a heterologous antibody. The latter antibody is directed against CudA, a nuclear protein that appears not to be regulated by DIF-1 (Fukuzawa et al., 1997). PCR was performed using primers directed against the ecmA promoter region. The promoter selected as a control was that of the G protein βsubunit gene (gpbA), another gene where there is no evidence for DIF-1 regulation. PCR conditions were set such that control genomic DNA gave approximately equal signals with the ecmA and gbpA PCR primers (Fig. 13).

The apparent abundances of the PCR products generated using the various chromatin samples are expressed relative to the value obtained using the gbpA PCR primers with chromatin from cAMP treated and DIF-induced cells, which was immunoprecipitated using the anti-DimB antibody. When the chromatin sample from cAMP and DIF-treated cells was analysed using the ecmA promoter region, there was an enrichment of approximately fourfold (Fig. 13): a value that is in good accord with the induction values for ecmA itself.

There was no enrichment when cAMP and DIF-1 were omitted(Fig. 13) and a lower,statistically non-significant degree of enrichment when either DIF-1 alone or cAMP alone was used (data not shown). Enrichment depended upon the presence of the DimB antibody; CudA antibody produced no enrichment(Fig. 13). In addition, there was no enrichment when dimB- cells were treated with cAMP and DIF-1(Fig. 13).

DimB was isolated by virtue of its binding to two spatially separated regions of the promoter of the ecmA gene, R1 and R2. Mutational analyses of the DimB-binding sites within R1 and R2 identified a region that is required for binding and this allowed a consensus for the binding sequence to be derived and confirmed (Fig. 5B). It is not a dyad and does not contain the ACGT `core'sequence that is found in the binding sites of many other bZIP proteins. However, the bZIP family shows a broad range of DNA binding specificities(Fujii et al., 2000), the sites are not always perfect dyads and the binding sites of several well characterised bZIP proteins, e.g. AP1 and GCN4(Fig. 5B), also lack an ACGT sequence.

The region within R1 that is essential for DimB binding is contained within the sequence CATCATC. Interestingly, a very similar sequence within the promoter of the 7E gene, CATCACC, is essential for expression in pstA cells and for repressing prestalk expression in prespore cells(Seager et al., 2001). However, we did not analyse the R1 element functionally. Instead we determined the biological function of the DimB binding site in R2.

As a recipient for cloning R2, and its mutant form R2pM1, a new construct,S, was created and this provided an unexpected result. Construct S contains an ecmA promoter fragment that is 38 nucleotides shorter at its distal end than construct O (Early et al.,1993). It also differs significantly from construct O at its cap-site promoter end, in that it employs ecmA basal transcription signals (Fig. 1, Fig. 6B). The latter fact probably explains why it is better expressed in pstA cells than construct O.

Although it has become the standard pstA marker, construct O is weakly expressed and is not totally specific; at extended times of staining it displays a finite level of expression in pstO cells and ALC. By contrast,construct S is highly specific to the pstA cells. This is manifest in a highly foreshortened region of prestalk staining and in the almost complete lack of expression in ALC. This expression pattern is, to our knowledge, novel. However, it is in accord with a previous study that suggested that the pstO cells and the ecmO:lacZ expressing ALC may be a unitary population(Abe et al., 1994). The fact that construct S is efficiently and selectively expressed in pstA cells make it a useful marker for future studies.

Construct R2S has, at its cap-site distal end, a very similar structure to construct O but, presumably because of its different cap-site proximal architecture, it is expressed at a much higher level and strong staining is also observed in pstO cells and in ALC. By contrast, R2pM1S has the same,highly pstA-specific, expression pattern as construct S [the construct that lacks R2 (Fig. 6B)]. Thus, the four point mutations that abrogate binding of DimB to R2 in vitro prevent expression in the pstO and ALC. Therefore, within the context of the promoter sequences contained within R2S, the -510 site is essential for pstO and ALC gene expression but not for pstA-specific expression.

There is evidence that DIF-1 selectively induces pstO-specific gene expression (Thompson et al.,2004; Thompson and Kay,2000b). Hence, the fact that pstO-specific expression is eliminated by the four point mutations present in R2pM1S implies a role for the -510 site in DIF-1 signalling. The corollary, that pstA expression is unaffected by the four point mutations present in R2pM1S, is also in accord with analyses of the DIF-1 deficient mutant dmtA-, which suggest that pstA-specific gene expression uses a DIF-1-independent signalling pathway(Thompson and Kay, 2000a).

Analysis of the dimB-null mutant shows that a prespore marker and the standard prestalk markers are expressed in a similar spatial pattern to the control. The most telling result is the presence of an apparently normal pstO region in the dimB-null strain. This is an important difference from both the dmtA- and dimB- mutants, where there is a defect in pstO cell differentiation. The dimB null does, however, show a more subtle difference from the control.

In control slugs, there is both an anteroposterior and a radial gradient of ecmA expression, with their respective peaks at the tip and the slug exterior; this creates a thimble shaped cell grouping that ensheaths the prestalk region. These gradients are lost in the dimB- strain. Hence,DimB functions as a repressor that reduces ecmA expression in the more posteriorly and centrally located prestalk cells.

A similar ecmA:lacZ staining pattern has been described previously, again for slugs migrating towards a dim unidirectional light source(Jermyn and Williams, 1991). Interestingly, the pattern was absent when slugs were allowed to migrate under overhead light (Jermyn and Williams,1991). Perhaps, therefore, exposure to a high light intensity downregulates the repressive activity of DimB and disrupts the ecmAexpression gradient.

EcmA is a major component of the slime sheath, the extracellular matrix that surrounds the slug and a DimB activity gradient may help ensure that EcmA is selectively synthesised in the region of the slug where it needs to be most highly concentrated: i.e. at the surface. Slug movement is directed by the tip, so if other genes needed for slug migration are modulated in a similar manner to ecmA, disruption of their graded expression could account for the lack of slug migration in the dimB- mutant.

It seems paradoxical that: (1) DimB should bind to the -510 site in vitro;(2) mutation of the -510 site should prevent ecmA expression in pstO and ALC; yet (3) the dimB- mutant should overexpress ecmA in parts of the prestalk region. We believe that this can best be explained if the -510 site is bound in vivo by two proteins: an, as yet unidentified,activator; and DimB, functioning as a repressor. The relative occupancy of the-510 site by these two proteins would establish the observed gradients of expression in the prestalk region. One obvious candidate for the activator is the previously mentioned MYB domain protein that binds to R2 but not R1 (M.F. and J.G.W., unpublished).

Analysis of the dimB- mutant provides a further apparent link between DimB and DIF-1 signalling: the mutant is DIF-1 non-responsive in a monolayer assay that measures induction of ecmA gene expression. However, ecmA and cell-type specific markers derived from its promoter are expressed during multicellular development of the dimB-strain.

The monolayer result presumably reflects some limitation of the assay system: deprived of cell-cell interactions and cell-matrix interactions, the signalling inputs to a monolayer cell would be expected to be significantly aberrant. The result can be explained by assuming that under monolayer conditions dimB acts alone, either as a direct activator of ecmA or as a factor needed to achieve DIF responsiveness. It implies a second protein, which is conditionally redundant with DimB, that receives signalling inputs from other cells and that fulfils these functions in normal development. Nineteen bZIP proteins are encoded within the Dictyostelium genome (Eichinger et al., 2005). Hence, there are abundant candidates for such a role. The crucial issue is whether DIF-1 is the normal inducer in the whole organism and here the biochemical evidence is very telling.

Although the genetic evidence from monolayer cells appears to be misleading, certainly as far as an obligate role for DimB in ecmAexpression, analysis of the behaviour of DimB establishes a clear link to DIF-1 signalling. DimB rapidly accumulates in the nucleus when cells are treated with DIF-1 and it becomes associated with the ecmA promoter after cells are exposed to a combination of cAMP and DIF-1. Many post-aggregative functions require that cells be continuously exposed to a high concentration of cAMP (Kay,1982; Mehdy et al.,1983; Schaap and van Driel,1985; Berks and Kay, 1988). Hence, we favour the notion that cAMP is also permissive rather than instructive in this instance.

Thus, we propose that DimB is induced by DIF-1 to accumulate in the nuclei of a subset of the prestalk cells and that, by binding to the ecmApromoter at the -510 site, DimB represses ecmA gene transcription. How then is DimB activated by DIF-1? Dd-STATc provides a precedent for DIF-1 induced nuclear accumulation of a Dictyostelium protein. However, in contrast to the STAT proteins, where there is a paradigm for activation, bZIP proteins display varied activation mechanisms. Some, such as the mammalian CREB protein, are constitutively nuclear and are activated as transcription factors by serine or threonine phosphorylation(Montminy, 1997). Others, such as the fission yeast protein Pap1, are regulated at the level of nuclear export (Toone et al., 1998). There is, to our knowledge, no precedent for activation of a bZIP by a direct tyrosine phosphorylation event of the kind that initiates the nuclear accumulation of DdSTATc. Presumably, therefore, two wholly or partially separate DIF-1 regulated signalling pathways activate these two different transcription factors.

During the course of the above study, we learnt that the same gene was under investigation by Huang et al. (Huang et al., 2005). These authors identified DimB bio-informatically,rather than biochemically, and they showed that it dimerises with DimA. The two studies are in generally good accord and the differences in ecmAexpression patterns that are observed in the slug are probably explicable by strain differences; they used strain Ax4, whereas we used Ax2.

This work was supported by Wellcome Trust Program Grant 053640/Z to J.G.W. We also thank Susan Ross for excellent technical assistance and Dr Jonathan Chubb for advice on the ChIP assay.