The apoptosis pathway appears to be evolutionarily conserved, both phenotypically and molecularly, across all metazoans studied to date, indicating that its basic features were likely present in the last common metazoan ancestor.6, 24
While the triggers for apoptosis vary depending on species, cell type, and environmental conditions, all apoptotic pathways in all species studied lead eventually to caspase activation, followed by dismantling of the apoptotic cell in a non-inflammatory manner. Here, we begin to globally determine apoptotic caspase cleavage sites in three non-human model metazoans and investigate the evolutionary conservation of the common caspase stage of apoptosis.
Our human data set, collected over six years, contains 1444 protein targets; the mouse, Drosophila, and C. elegans data sets are roughly an order of magnitude smaller, reflecting a smaller number of experiments on these species. Furthermore, eight different human cell lines were examined, compared with three mouse cell lines and one Drosophila cell line. Despite the incomplete sampling of the non-human metazoans, we can make meaningful positive comparisons between these smaller sets against the much more complete human data set because the low false positive rate (<3%, data not shown) provides high confidence identification.
All comparative proteomics studies come with a risk of bias towards high-abundance proteins. High abundance tends to be positively correlated with high conservation,28
so sampling bias towards high abundance proteins might yield a false result of high conservation. However, our experimental technique reduces the impact of this problem. By enriching samples at the tryptic peptide level (leaving only one or a few peptides per protein), our N-terminal labeling method decreases the complexity of the peptide mixture further than traditional methods, diminishing the chance of low abundance proteins being masked by extremely high abundance proteins like actin. In fact, studies using our method in human serum detected proteins with known abundance values spanning six orders of magnitude.29
Using the PaxDB database,12, 13
we have established high coverage of the abundance range of the caspase substrates identified in our cell line-based human reference data set (). Therefore, we are confident that abundance bias does not have a large role in our results.
Our results suggest a hierarchical arrangement of conservation, depicted in . Caspases target a specific set of pathways across metazoans in order to complete apoptosis. In some cases, caspases in different species target different proteins within the same pathways. In other cases, the same protein remains a caspase substrate across a broad lineage, and in still other cases, caspases even target the same cleavage site motif in different species. Our data are consistent with previous observations that short linear motifs are poorly conserved over evolution,8
but also show that loss of the short linear motif recognized by caspases is not likely to stop the same protein from being targeted.
Figure 5 Results of this study show that caspases tend to recognize and cleave particular motifs only across short evolutionary distances (represented by the human–mouse comparison, spanning <100 million years of evolution37), but that the same (more ...)
Our study has yielded similar conservation patterns to those observed for phosphorylation and transcription factors. Tan et al.9
used mass spectrometry to identify phosphorylation sites in Drosophila
, C. elegans
, and yeast, and compared them with known human sites. They found that the set of modifications conserved at the motif level was small, but networks defining relationships between kinases and substrates were conserved more broadly. Additional computational phosphorylation studies have also revealed that motif-level conservation is weaker than protein-level conservation.30, 31
Studies of transcription factor binding in yeast show a somewhat similar pattern: while DNA binding sites are only modestly conserved, overall regulatory networks and the functions they impact are retained across hundreds of millions of years of evolution.32, 33, 34
Notably, three studies of single Drosophila
caspase substrates have also demonstrated protein level (but not motif level) conservation.35, 36, 37
Caspases, like all enzymes that catalyze specific post-translational modifications, face the challenge of recognizing their particular targets in the complex cellular milieu. How caspases achieve this is not well understood. Executioner caspases, like human caspase-3, are functionally polyspecific, meaning that they recognize several otherwise unrelated substrates, and yet are restrained from targeting all Asp sites in all proteins in a digestive manner.6
The degeneracy of the physiological cleavage site consensus sequence seen in all four organisms (), plus the higher conservation at the protein level than the motif level, suggest that other factors, such as exosite binding,38, 39
may influence what is cleaved in cells.
We examined UniProtKB, Flybase,40
and Wormbase (www.wormbase.org
, release WS229) annotations for the 62 Drosophila
and 19 C. elegans
caspase substrate proteins that are specific to those species, having either no human orthologs, or human ortholog(s) not known to be substrates. Many have little or no functional annotation, but several in both species are either known or predicted to be involved in development (). In Drosophila
, three of these proteins function in spermatogenesis, three in neurogenesis, and three in other types of development. Five C. elegans
-specific proteins are either known or predicted to be involved in embryo development. Apoptosis is an important aspect of development in all species, but further studies are needed to determine whether these cleavages are relevant to developmental apoptosis. Three of the five C. elegans
development proteins are known to be extracellular. It is possible that the caspase cleavages of these proteins are not physiologically relevant, but result from the C. elegans
extract being made from whole bodies, rather than a cell line.
Subset of Drosophila and C. elegans Caspase Substrates with no Human Ortholog
Another interesting finding is that three proteins cleaved in Drosophila
are all members of the gypsy chromatin insulator complex. This complex regulates accessibility of certain chromosomal regions,41
so these cleavages are in accordance with the GO analysis showing the tendency for caspases to cleave chromatin-associated complexes (Supplementary Table 3
In aggregate, these studies represent a unique systematic comparison of apoptotic caspase substrates across metazoans, shedding light on what is functionally most important. In all three non-human species, at least half of the substrates identified are conserved with human substrates on either the motif, protein, or pathway level, and given that our human caspase substrate data set is likely not comprehensive, this could be an underestimate. This finding supports the view that a substantial fraction of caspase cleavages do serve important functions that have been conserved by natural selection across 600 million years of metazoan evolution.