Prokaryotic caspase homologs: phylogenetic patterns and functional characteristics reveal considerable diversity

Abstract

Caspases accomplish initiation and execution of apoptosis, a programmed cell death process specific to metazoans. The existence of prokaryotic caspase homologs, termed metacaspases, has been known for slightly more than a decade. Despite their potential connection to the evolution of programmed cell death in eukaryotes, the phylogenetic distribution and functions of these prokaryotic metacaspase sequences are largely uncharted, while a few experiments imply involvement in programmed cell death. Aiming at providing a more detailed picture of prokaryotic caspase homologs, we applied a computational approach based on Hidden Markov Model search profiles to identify and functionally characterize putative metacaspases in bacterial and archaeal genomes. Out of the total of 1463 analyzed genomes, merely 267 (18%) were identified to contain putative metacaspases, but their taxonomic distribution included most prokaryotic phyla and a few archaea (Euryarchaeota). Metacaspases were particularly abundant in Alphaproteobacteria, Deltaproteobacteria and Cyanobacteria, which harbor many morphologically and developmentally complex organisms, and a distinct correlation was found between abundance and phenotypic complexity in Cyanobacteria. Notably, Bacillus subtilis and Escherichia coli, known to undergo genetically regulated autolysis, lacked metacaspases. Pfam domain architecture analysis combined with operon identification revealed rich and varied configurations among the metacaspase sequences. These imply roles in programmed cell death, but also e.g. in signaling, various enzymatic activities and protein modification. Together our data show a wide and scattered distribution of caspase homologs in prokaryotes with structurally and functionally diverse sub-groups, and with a potentially intriguing evolutionary role. These features will help delineate future characterizations of death pathways in prokaryotes.

Figure 1. Taxonomic distribution of putative prokaryotic metacaspases in sequenced genomes.

The mean metacaspase abundance (metacaspases per 1000 proteins) is shown in the upper bars (light blue), for each group (i.e. phylum, except for Proteobacteria, which have been divided into classes, unclassified Bacteria, which encompasses Candidatus Cloacamonas acidaminovorans and Thermobaculum terrenum ATCC BAA-798, and unclassified Proteobacteria, harboring the single member Magnetococcus sp. MC-1). The lower bars show the total number of sequenced genomes within each group. These are divided into categories based on the range of their metacaspase abundance (metacaspases per 1000 proteins). The prokaryotic groups have been ordered on mean metacaspase abundance, with the highest on top. The “Other” categories comprise groups that are represented by five or less sequenced genomes and lack putative metacaspases. Scaling is the same for both the bacterial and archaeal charts.

Figure 2. Prokaryotic metacaspase domain architectures and predicted functions.

Domains and distances between domains are not drawn to scale, but the N- to C-terminal order is the same in all sequences sharing a particular domain architecture. Consecutively repeated domains have been grouped, indicated by “repeat” following the domain name. The number of repeats can differ between the sequences within one domain architecture group. “iTMHo” is a transmembrane helix with N-terminal domains predicted to be situated intracellularly and C-terminal domains extracellularly. The topology is the reverse for “oTMHi”. N-terminal transmembrane helices are likely to be signal peptides. (A) Metacaspase architectures ordered by the number of different prokaryotic groups in which they are found. The total number of sequences displaying each domain architecture is also shown. The colors reflect the function(s) of each domain and are defined in B. (B) Functional categories and color key of domains detected in addition to the metacaspase domain, ordered on the number of sequences in which they occur. Note that a single sequence, or a single domain, may have several of these functions. The function categories are based on the functional information that can be obtained from Pfam (http://pfam.sanger.ac.uk/) for each domain type. (C) Domain architectures which are unique to Cyanobacteria, and would fall into the lowermost (“Other”) category in A. Domain types which are only found in Cyanobacterial metacaspases (but may be found in other proteins in other organisms) have been emphasized by a red outline.

Figure 3. Phylogenetic distribution of Cyanobacterial metacaspases.

Cyanobacterial phylogenetic tree based on 285 core orthologs, after . The number of identified metacaspases have been plotted for each genome that was analyzed. The tree also shows some Cyanobacteria with draft genomes only, which were not included in this study (shaded). The clade encompassing 26 members of the unicellular genera Synechococcus, Cyanobium and Prochlorococcus has been collapsed as none contained metacaspases. The occurrence of the traits nitrogen fixation, filamentous morphology and symbiotic competence is shown in the figure. Letters within brackets, e.g. “(a)”, are references to the detailed domain architectures in Fig. 2.

Figure 4. Phylogenetic relationship of polysaccharide deacetylase-containing prokaryotic metacaspases.

A set of highly conserved metacaspase sequences found in the Gammaproteobacterial genera Xyllela, Stenotrophomonas and Xanthomonas, including Xanthomonas campestris, in which the protein has been implicated in programmed cell death (PCD) , , as well as in the Acidobacteria Candidatus Solibacter usitatus Ellin6067 and Candidatus Koribacter versatilis Ellin345. The metacaspase involved in Xanthomonas campestris pv. glycines PCD (GI 88792532) is emphasized in red. The metacaspase domains of all 17 sequences identified in genomes analyzed in this study have conserved cysteine-histidine catalytic dyads. The maximum likelihood phylogenetic tree to the left was produced by alignment of the sequences with Muscle v3.8.31 followed by tree construction with FastTree v2.1.3 . Three cysteine-histidine positive sequences from Trichodesmium erythraeum IMS101 (Cyanobacteria), Pseudomonas syringae pv. tomato str. DC3000 (Gammaproteobacteria) and Pyrocoocus yayanosii CH1 (Euryarchaeota) were used as an outgroup.

Figure 5. Examples of operons with metacaspase genes.

Operon structures were obtained from the MicrobesOnline operon database (http://www.microbesonline.org/operons/, [47], [48]), supplemented with genomic region data from NCBI (http://www.ncbi.nlm.nih.gov/gene). The numbers to the left and right of each operon are the start and end nucleotide positions. The start is higher than the end when the operon is situated on the minus strand. GenBank accession IDs are shown at the beginning of each gene. The protein products of the Trichodesmium erythraeum IMS101 and Xanthomonas campestris pv. campestris str. 8004 metacaspase genes are displayed in Fig. 2C and Fig. 4, respectively.