Spatio-temporal expression patterns of P. pallidum acaA homologs
In D. discoideum (Ddis), acaA shows complex expression from different promoters. The promoter proximal to the coding sequence directs expression high expression at the slug tip, the central promoter directs low expression in the prespore region, while the most distal promoter directs high expression during aggregation [15]. P. pallidum (Ppal) has three acaA genes, aca1, aca2 and aca3 (Additional file 1, Figure A1). Comparative transcriptomics shows that these and other acaA genes across taxon groups are upregulated after starvation, with group 4 acaA genes showing peak expression during aggregation. Aca genes are most highly expressed in stalk cells in groups 1–3, but in group 4 expression is highest in cup cells, which are unique to group 4 (Fig. A1).
To investigate the spatial expression pattern of Ppal aca1, aca2 and aca3, their promoter regions were fused to the LacZ reporter gene and transformed into Ppal cells. Developing structures were fixed and incubated with X-gal to visualize β-galactosidase activity. Aca1 was not expressed during aggregation and started to be expressed weakly at the utmost tip region of the primary sorogen, and later sometimes in the tip of secondary sorogens (Fig. 1A). Aca2 and aca3 were already expressed in aggregates and more strongly during post-aggregative development (Fig. 1B,C). In primary sorogens, aca2 was expressed throughout the structure but most strongly at the tip region. Aca3 expression was more specific to the tips of primary and secondary sorogens. Overall, the post-aggregative expression pattern of the three Ppal acas resembles that of Ddis acaA with strongest expression at the sorogen tips [15, 16].
Deletion of aca genes in P. pallidum
To assess the biological roles of the three Ppal aca genes, we replaced essential regions in each gene with the LoxP-NeoR cassette, in which NeoR, the single selectable marker of Ppal is flanked by loxP sites (Additional file1, Figure A2). The aca1ˉ clones aggregated normally and formed fruiting bodies with somewhat thinner and longer stalks than those of wild type Ppal (Fig. 2A, D). The aca2ˉ mutant aggregated and formed the primary sorogen normally but showed delayed formation of the first whorls of secondary sorogens (Fig. 2C). Such whorls arise at regular intervals when a posterior segment of the primary sorogen pinches off, while forming several regularly spaced tips, which each give rise to a small side branch. As a result, the branch-less lower stalk of aca2- fruiting bodies was longer than in wild-type (Fig. 2D). The aca3ˉ mutant formed few aggregation centres with long streams (Fig. 2A), that partitioned into many tip-forming small aggregates that each gave rise to a small fruiting body. The central, large aca3- aggregate produced a normal fruiting body (Fig. 2B,D). Overall, the phenotypes of single aca knock-out mutants were subtle. We tried to generate double and triple aca knock-outs by recycling the LoxP-NeoR cassette using the cre-recombinase expression vector pA15NLS.Cre [17]. This succeeded for the aca1ˉ mutant, allowing us to generate aca1ˉaca2ˉ and aca1ˉaca3ˉ double knock-outs, but not for the aca2ˉ or aca3ˉ knock-outs. The aca1ˉaca3ˉ phenotype combined features of aca1ˉ and aca3ˉ knock-out mutants. Similar to aca3ˉ, aca1ˉaca3ˉ formed few but large streaming aggregates (Fig. 2A,B), while the fruiting bodies showed thinner and longer stalks, like the aca1ˉ mutants (Fig. 2D). The aca1ˉaca2ˉ cells aggregated and formed primary sorogens normally. However, the separation of the first whorl only occurred after 28 h of starvation, when WT, aca1ˉ, and aca2ˉ had already formed fruiting bodies (Fig. 2C). As a result, aca1ˉaca2ˉ made very tall fruiting bodies with side branches only at the upper stalk (Fig. 2D).
The failure to recycle LoxP-NeoR cassette of aca2ˉ or aca3ˉ mutants was probably due to limited selectability of cells transformed with pA15NLS.Cre with its G418 selection cassette. We found that Ppal growth is also inhibited by the antibiotic Nourseothricin. This allowed us to use a Cre-recombinase expression vector pDM1483 [18] with a Nourseotricin selection cassette to eliminate LoxP-NeoR from aca3ˉ and aca1ˉaca3ˉ and to generate aca3ˉaca2ˉ and aca1ˉaca3ˉaca2ˉ knock-outs.
Compared to wild-type, aca1ˉaca2ˉ, aca1ˉaca3ˉ and single aca knock-outs, which all initiated aggregation within 8 h of starvation, the aca3ˉaca2ˉ mutant only started to aggregate at 24 h or later (Fig. 3A). Only few aggregation foci were formed, which attracted very long aggregation streams. Starting from the initial (small) focus, mounds appeared at intervals within the streams, which each attracted downstream cells. Each of these mounds gave rise to a small, branched fruiting body, which, similar to aca2ˉ, showed a longer whorl-free lower stalk (Fig. 3B). The aca1ˉaca3ˉaca2ˉ phenotype combined features of the aca1ˉaca2ˉ and the aca3ˉaca2ˉ mutant. Similar to aca3ˉaca2ˉ, aggregation was much delayed with long streams appearing only after 24–48 h of starvation (Fig. 3A), which eventually broke up and gave rise to small fruiting bodies. These fruiting bodies showed delayed side-branch formation, like aca1ˉaca2ˉ (Fig. 3B). Staining of the aca1ˉaca3ˉaca2ˉ stalk and spore cells with the cellulose dye Calcofluor showed that it formed a normal primary and secondary stalk and elliptical spores encapsulated in cellulose walls (Fig. 3C), and this was also the case for all other acaˉ mutants (not shown).
To investigate whether the aggregation phenotypes of the aca3ˉaca2ˉ or aca1ˉaca3ˉaca2ˉ mutants were cell-autonomous, the mutants were developed as chimeras with wild-type cells. Introduction of 10% wild-type cells was sufficient to restore delayed aggregation of both mutants (Fig. 3A). The mixtures aggregated within 8 h of starvation like wild-type cells, but still formed larger aggregation streams. Also, the formation of secondary sorogens in aca1ˉaca3ˉaca2ˉ chimeras with wild-type was not as delayed as in aca1ˉaca3ˉaca2ˉ alone, resulting in formation of more normal fruiting bodies (Fig. 3B). These experiments show that the defects in aggregation and whorl separation of the acaˉ mutants are non-cell autonomous.
We also tested microcyst formation in aca knock-out mutants. Incubation with 0.2 M sorbitol for 24 h induced cyst formation in both wild type and aca1ˉaca3ˉaca2ˉ to the same degree (Additional File1, Figure A3), indicating that the aca genes are not required for encystation.
Restoration of aca1ˉaca3ˉaca2ˉ aggregation by 8Br-cAMP.
The strongly reduced initiation of aggregation centres and extensive delay in aggregation of both the aca3ˉaca2ˉ and aca1ˉaca3ˉaca2ˉ was unexpected, since Ppal does not use cAMP as attractant for aggregation, but most likely glorin [5, 9]. However, both Ddis and Ppal require PKA activity and therefore likely intracellular cAMP to develop competence for aggregation [19, 20]. To investigate whether lack of PKA activation due to the absence of intracellular cAMP cause the aggregation abnormalities in aca3ˉaca2ˉ and aca1ˉaca3ˉaca2ˉ, aca1ˉaca3ˉaca2ˉ cells were developed on agar containing 2.5 mM 8Br-cAMP, a membrane-permeant PKA agonist. While without 8Br-cAMP cells had not yet started to aggregate after 24 h of starvation, the 8Br-cAMP treated cells initiated many aggregation centres and almost completed aggregation within 6 h (Fig. 4). The aggregates remained however blocked in the mound stage and did not form fruiting bodies. This was however also the case for most Ppal WT aggregates developed on 8Br-cAMP agar. These results show that the aca1ˉaca3ˉaca2ˉ aggregation defect was likely caused by insufficient intracellular cAMP for PKA activation.
cAMP relay in Ppal and in Ddis aca-/ACG cells
Despite the loss of all ACA activity the aca1ˉaca3ˉaca2ˉ cells still made relatively normal fruiting bodies after a long delay. cAMP-induced excitation and adaptation of ACA underpins pulsatile cAMP signalling and wave propagation in Ddis [21, 22], with cAMP receptors (cARs) and extracellular cAMP phosphodiesterase (PdsA) as essential components to respectively detect secreted cAMP and to hydrolyse it between pulses [23–25]. From earlier findings that cAR or pdsA null mutants in Ppal were specifically defective in fruiting body morphogenesis [11, 12], we concluded that cAMP waves mediated this process as they do in Ddis [4]. The present data imply that this is either not the case, or that the aca1ˉaca3ˉaca2ˉ cells have a means to compensate for loss of ACA activity.
To investigate whether Ppal also shows transient cAMP induced cAR mediated accumulation of cAMP, we stimulated wild-type Ppal cells at different stages of development with the cAR agonist 2'H-cAMP in the presence of the PdsA inhibitor DTT [26]. Figure 5A shows that cells at all stages show a basal level of 3–6 pmoles cAMP/mg protein. Starving cells or cells from streaming aggregates showed none or marginal responses to 2'H-cAMP, while cells from tipped mounds showed a 5 pmoles/mg protein increase in cAMP, which then levelled off. However, cells from dissociated sorogens showed transient increase that peaked after 3 min after stimulation at 11 pmoles above basal levels and then decreased to 5 pmoles. These data indicate that Ppal can relay a pulse of cAMP, but only after tips and sorogens have formed. In Ddis, which unlike Ppal also uses cAMP to aggregate, cAMP relay is highest at the aggregation stage [27]. We could not meaningfully measure 2'H-cAMP induced cAMP accumulation in the aca1ˉaca3ˉaca2ˉ cells, because only few aggregates are formed at different times, which then fragment and fairly rapidly mature into fruiting bodies. This means that at any time only a very small fraction of cells is in the sorogen stage.
Apart from Aca1, Aca2 and Aca3, two other adenylate cyclases, AcgA and AcrA are expressed in Ppal sorogens [28]. The experiment in Fig. 5A does not identify the adenylate cyclase responsible for the cAMP increase. While currently not feasible in Ppal, a Ddis acaA knock-out is available that expresses AcgA (ACG) from the constitutive actin 15 promoter [29]. During growth, this mutant synthesizes cAMP at a constant rate [30], but it is unknown whether cAMP synthesis comes under cAR regulation at a later stage. We compared 2'H-cAMP induced cAMP accumulation between Ddis wild-type and acaˉ/ACG cells in vegetative and 4 h starved cells, which are just starting to aggregate. To validate that the observed responses are mediated by Ddis cAR1, we included the cAR1 antagonist 2'3'-O-isopropylidene adenosine (IPA) in control assays. Figure 5B shows that wild-type Ddis shows no 2'H-cAMP induced cAMP accumulation in the vegetative stage and a 70 pmoles/mg protein increase in 4 h starved cells that peaks at 5 min. This response is almost completely inhibited by IPA. Vegetative acaˉ/ACG cells show a steady increase in cAMP levels after addition of 2'H-cAMP/DTT that is only slightly reduced in the presence of IPA. However, 4 h starved acaˉ/ACG cells show a faster transient increase of cAMP that peaks at 3 min after 2'H-cAMP/DTT stimulation. This response is also strongly reduced by IPA. These data indicate that in early aggregating Ddis cells, ACG is also controlled by cAR stimulation. The apparent ability of other adenylate cyclases than ACA to participate in transient cAR mediated cAMP accumulation provides some resolution for the contrasting effects on Ppal fruiting body morphogenesis of aca deletion on one hand and cAR or pdsA deletion on the other.