Fig 1.
Histidine-aspartate phosphotransfer architectures.
(A) A canonical two-component signaling system consists of a histidine kinase and a response regulator, wherein a signal is transmitted by transfer of a phosphoryl group from a conserved histidine in the HisKA domain (teal oval) to a conserved aspartate in the REC domain in the response regulator (red rectangle). (B) The B. subtilis Spo0 pathway is a phosphorelay. Signal transduction is initiated by activation of one of the five sensor histidine kinases that are associated with this pathway. The phosphoryl group is transferred from the HisKA domain in the kinase to the Spo0F REC domain (blue rectangle), then to the phosphotransferase Spo0B (orange oval), and finally to the REC domain of Spo0A (green rectangle). Spo0F lacks an output domain; Spo0A has the domain architecture of a typical response regulator, including a REC domain and a DNA-binding domain. (C) The C. acetobutylicum Spo0 pathway has a direct phosphorylation architecture, wherein multiple sporulation kinases are capable of direct transfer of a phosphoryl group to Spo0A.
Fig 2.
Signaling specificity and crosstalk avoidance.
(A) Molecular recognition maintains signaling fidelity between cognate histidine kinase–response regulator pairs and prevents phosphotransfer between non-cognate proteins encoded within the same genome (e.g., two-component signaling systems shown at left). Each response regulator is capable of recognizing multiple histidine kinase specificity signatures. The set of kinase specificity signatures recognized by the response regulator is represented qualitatively as a spectrum (right). Selection likely acts to separate the specificity spectra of response regulators encoded within the same genome, resulting in little or no overlap between spectra. Each histidine kinase must occupy a non-overlapping region of the specificity spectrum of its cognate response regulator. (B) The requirements of signaling fidelity exert greater constraints on the specificity signatures of a phosphorelay. The phosphorelay sporulation kinase (HK) must interact with Spo0F and not Spo0A, while Spo0B must interact with both Spo0F and Spo0A (left). The phosphorelay interaction pattern requires that the spectra (right) for Spo0F (blue) and Spo0A (green) must overlap and that Spo0B (orange dot) be located in the overlapping region. Additionally, sporulation kinases (e.g., teal dot) must be located in the Spo0F specificity spectrum, but outside of the region that overlaps with the Spo0A spectrum.
Fig 3.
Phylogenetic distribution of predicted Spo0 pathway proteins.
Cladogram of Firmicutes species used in this study, annotated with colored dots indicating predicted Spo0 pathway proteins: one or more orphan kinases (cyan); Spo0F (blue); Spo0B (orange); Spo0A (green). A filled cyan dot indicates a genome that encodes at least one orphan kinase with a PAS domain. The number of orphan kinases in each genome is given in S5 Table. Stars indicate genomes used in the in vitro phosphotransfer assays reported in this study. Phylogeny constructed from the concatenated alignments of 50 ribosomal protein families using RAxML with the CAT model and 100 bootstrap replicates (branch support values greater than or equal to than 50 are shown). Colored branches indicate species that are known to sporulate in Class Bacilli (blue) and Class Clostridia (red) (see also S1 Table). Species in which sporulation has not been reported are shown in grey. Tree representation created using ITOL [71]. See S1 Fig for the corresponding phylogram showing the outgroup used to root the tree and support values on all branches.
Table 1.
Predicted specificity residues in B. subtilis, D. acetoxidans, and C. acetobutylicum Spo0 proteins.
Table 2.
Protein constructs.
Fig 4.
Phosphotransfer profiling of Spo0 phosphorelay proteins from Desulfotomaculum acetoxidans.
All eight combinations of Dtox1918 and one or more downstream constituents of the predicted Dtox phosphorelay were examined for phosphotransfer (see text for details). The observed interactions are consistent with a Spo0 phosphorelay in D. acetoxidans: Spo0F and Spo0B are both necessary and sufficient for phosphorylation of Spo0A. Further, phosphotransfer to Spo0B was only observed in the presence of Spo0F. (A) The proteins in each reaction (listed above the corresponding lane) were added sequentially at 4 minute intervals (time points noted at right). Reactions were sampled 3 minutes after addition of the final constituent protein. (B) Reactions from (A) that included Spo0A were sampled again 10 minutes after addition of Spo0A. Direct phosphorylation of Spo0A was not observed even following this longer incubation period. See Table 2 for abbreviations.
Fig 5.
Specificity residues in predicted Spo0 architectures.
Sequence logos for predicted specificity residues of orphan kinases, Spo0F, Spo0B, and Spo0A in (A) Bacillar phosphorelays, (B) Clostridial phosphorelays, and (C) direct phosphorylation architectures. Created using WebLogo [72]. (D) Clade level summary of the phylogenetic distribution of predicted Spo0 proteins and architectures in spore-forming Firmicutes. Only spore-forming clades shown. Colored branches indicate predicted Spo0 phosphorelay (black) or direct phosphorylation (grey) architecture. Colored circles as in Fig 3 (E) Specificity residues of sporulation kinases, experimentally verified in this or prior studies (see also S1 Table), grouped by architecture.
Fig 6.
Cross-species complementation of B. subtilis Spo0 phosphorelay with D. acetoxidans Spo0 proteins.
Phosphotransfer was examined at each transition in the B. subtilis phosphorelay by systematic replacement of each B. subtilis protein with its D. acetoxidans counterpart (lanes 1–6). For comparison, phosphotransfer was also examined in ensembles of proteins lacking one or more constituents of the phosphorelay (lanes 7–10). Consistent with evolutionary conservation of phosphorelay interaction specificity, the D. acetoxidans phosphorelay protein rescued the phosphodonor and phosphoreceiver functions of the corresponding B. subtilis phosphorelay protein in each reaction. Phosphorylation of D. acetoxidans Spo0A by B. subtilis KinA was not observed, except in the presence of both Spo0F and Spo0B. In each reaction, following autophosphorylation, all proteins combined were incubated for 5 minutes (see text for details). Note that two bands were observed for Bs0B, the lower band likely corresponds to Bs0B that has lost its affinity tag. See Table 2 for abbreviations.
Fig 7.
The contributions of change and conservation in Spo0 protein specificity to pathway remodeling.
Cross-species phosphotransfer profiling probes the relative positions of Spo0 pathway specificity spectra across architectures. Horizontal axis represents phosphodonor specificity signatures, as in Fig 2. Vertical lines connect points corresponding to the same phosphodonor. (A) Schematic showing the positions of receiver specificity spectra in interaction space under the hypothesis that pathway remodeling arose via changes in kinase specificity. The Spo0A specificity spectra are similarly positioned in both architectures. This hypothesis predicts that every kinase that directly phosphorylates Spo0A in its native environment (filled circles) will also be capable of heterologous phosphorylation of Spo0A proteins from phosphorelays (open circles). (B) Positions of receiver specificity spectra under the hypothesis that pathway remodeling arose via changes in the Spo0A specificity spectrum. This hypothesis predicts the following heterologous interactions (open circles): Every direct phosphorylation kinase interacts with Spo0F proteins from phosphorelays and every phosphorelay kinase interacts with Spo0A proteins associated with direct phosphorylation architectures. (C) Receiver specificity spectra in the B. subtilis phosphorelay, the D. acetoxidans phosphorelay, and the C. acetobutylicum direct phosphorylation architectures. The relative positions of these spectra were inferred from experimentally determined autologous (filled circles) and heterologous (crossed open circles) interactions (summarized in Table 3; see also Figs 6, 8 and 9). The observed interactions support the hypothesis that pathway remodeling arose primarily through changes in kinase specificity, as shown in (A), with a minor shift in the Spo0A specificity spectrum of D. acetoxidans.
Fig 8.
Cross-species phosphotransfer profiling of phosphorelay histidine kinases and Spo0B phosphotransferases.
Assessment of phosphotransfer from (A) B. subtilis Spo0B or (B) D. acetoxidans Spo0B to Spo0A proteins from C. acetobutylicum, D. acetoxidans, or B. subtilis, as indicated above each lane. In these reactions, both Bacillar (A) and Clostridial (B) Spo0B specificity residues enable phosphotransfer to Spo0A proteins from both phosphorelays and direct phosphorylation architectures. (C) Phosphotransfer was examined from B. subtilis KinA (lanes 1–5) or D. acetoxidans Dtox_1918 (lanes 6–10) to Spo0A or Spo0F proteins, as indicated. The observed interactions are consistent with conservation of Spo0A specificity, with minor shifts, across pathway architectures and taxonomic classes. In each reaction, following autophosphorylation, all proteins combined were incubated for 5 minutes (see text for details). See Table 2 for abbreviations.
Fig 9.
Cross-species phosphotransfer profiling of direct phosphorylation histidine kinases.
Assessment of phosphotransfer from C. acetobutylicum kinases (A) Ca_C0903 and (B) CA_C3319 to Spo0A proteins from C. acetobutylicum, D. acetoxidans, or B. subtilis, as indicated above each lane. (C) Examination of phosphotransfer from Ca_C0903 and CA_C3319 to Spo0F proteins from D. acetoxidans or B. subtilis, as indicated. Phosphotransfer to Spo0F proteins by the phosphorelay kinase B. subtilis KinA, shown for comparison (third panel). The observed interactions are consistent with pathway rewiring via changes in sporulation kinase specificity. In each reaction, following autophosphorylation, all proteins combined were incubated for 5 minutes. See text for details; abbreviations are provided in Table 2.
Table 3.
Cross-species phosphotransfer interactions.
Fig 10.
Hypotheses for the evolutionary history of Spo0 architectures.
Hypotheses are represented on Firmicute cladograms (only spore-forming clades are shown). Branch color denotes predicted ancestral (internal branches) or extant (leaves) Spo0 pathway architecture (gray: direct phosphorylation architecture; black: phosphorelay). (A) The ancestral direct phosphorylation architecture hypothesis [32–34], with emergence of the phosphorelay in the Bacillar ancestor, predicts that phosphorelays and direct phosphorylation architectures in present-day genomes will be restricted to Bacilli and Clostridia, respectively. This distribution is inconsistent with our findings, shown at right. (B) The ancestral phosphorelay hypothesis that we propose entails a single invention of the Spo0 phosphorelay, followed by multiple transitions from phosphorelay to direct phosphorylation architecture. Our results are most consistent with this hypothesis. (C) An ancestral direct phosphorylation architecture with the present-day Spo0 pathway distribution that is consistent with our findings. This hypothesis requires multiple independent inventions of a phosphorelay or multiple acquisitions of a phosphorelay by horizontal transfer (yellow arrows). We consider both of these scenarios to be unlikely (see text for details).
Fig 11.
Evolutionary remodeling of a phosphorelay to a direct phosphorylation architecture.
Candidate scenarios whereby direct phosphorylation of Spo0A is gained and phosphorelay interactions are lost. (A) Sporulation kinase K1 accrues substitutions, resulting in a change of specificity from Spo0F to Spo0A. Subsequently, Spo0F and Spo0B are lost. (B) A hybrid histidine kinase (HHK) harbors a kinase interaction domain that is specific for Spo0A, but interaction with Spo0A is blocked by its native REC domain [53]. Loss of this REC domain allows interaction with Spo0A. Subsequently, K1, Spo0F, and Spo0B are lost. (C) Cell1 harbors a Spo0 pathway; Cell2 does not. Cell2 possesses a kinase, K2, that is specific for the Spo0A in Cell1. Acquisition of K2 by horizontal gene transfer results in direct phosphorylation of Spo0A. Subsequently, K1, Spo0F, and Spo0B are lost. (D) Cell1 and Cell2 both harbor Spo0 pathways. The specificity spectra for the Spo0 phosphorelay of Cell2 are shifted relative to the Spo0 spectra of Cell1, such that sporulation kinase K2 is specific for the Spo0A in Cell1. Acquisition of K2 by horizontal gene transfer results in direct phosphorylation of Spo0A. Subsequently, K1, Spo0F, and Spo0B are lost.