This web page was produced as an assignment for an undergraduate course at Davidson College.

The Conservation of the Protein Interactome Across Taxa

Summary:
This paper introduces a new way to look at the evolutionary conservation of protein interactomes by studying the multiprotein complexes of a number of highly evolutionarily divergent species. The authors’ stated goal was to identify protein complexes and physical protein-protein interactions (PPIs) that are evolutionarily conserved across a wide range of phyla. They also studied the functional consequences of these PPIs. The authors analyzed protein complexes from five organisms: worms (Caenorhabditis elegans), fruit flies (Drosophila melanogaster), mice (Mus musculus), sea urchins (Strongylocentrotus purpuratus), and humans. They independently validated these data using analyses of frog (Xenopus laevis), sea anemone (Nematostella vectensis), amoeba (Dictyostelium discoideum), and yeast (Saccharomyces cerevisiae).

The authors used co-fractionation of the protein complexes to extract the complexes from the cells and separate them based on their biochemical properties via techniques such as high-performance liquid chromatography. In the next step of the experiment, the researchers used tandem mass spectrometry (MS/MS) to identify the compounds. The proteins were first ionized and then sorted based on charge in the first of two tandem MS steps. In the second MS step, the researchers were able to identify the proteins based on the peaks present in the MS spectra.

This study demonstrated a method that allows researchers to identify the evolutionary conservation of protein subunits and PPIs across phyla that have diverged as long ago as a billion years. The authors found that evolutionarily ancient complexes tend to be involved in key cellular processes, while complexes containing proteins that evolved more recently tend to have functions related to multicellularity and are mostly present in Metazoa. They were able to identify and functionally validate novel protein complexes, as well as showing the conservation of complex function in phyla where these functions had not previously been confirmed. Finally, they showed that their method is able to detect PPIs in highly complex signaling networks. These results have broad implications for our ability to investigate the interactome of a wide range of species.

Evaluation:
I was extremely impressed by the amount of work that went into this paper. It was more of a basic science paper than translational research, but as a scientist it was fascinating to see how the authors were able to detect multiprotein complexes and their evolutionary conservation. Additionally, the last figure showed that many of their predictions were functionally relevant. Therefore, this method could have important implications in nearly every field of biology since these PPIs are key to the function of cells and organisms. However, the authors didn’t explicitly state any of these implications. Rather, they left the reader to think of them. Therefore, I think it would have been helpful if they had, for example, further interpreted the importance of the results of Figure 5 to show their applications. Additionally, some of the figures were slightly confusing and the figure legends were not always very helpful in explaining them. My main issues were with the workflow in Figure 1, the protein complexes in Figure 3 (it was difficult to see how the different parts of the figure related to each other and this was not explicitly stated in the legend), and the way they presented their AP/MS data in Figure 4a.

Figures:

Figure 1: This figure shows the sequence of steps the authors went through to obtain their results. First, the figure shows the phylogenetic relationships between the organisms studied (panel A). Next, the authors display sample data from the co-fractionation experiments that separated protein complexes based on their biochemical properties as well as the MS/MS experiments that were used to identify the proteins based on their mass spectrometry profiles (panel B). The right-hand side of panel B also shows the outputs from these analyses. Finally, the authors show the computational analyses they performed to identify conserved complexes (panel C). Proteins that co-eluted (suggesting that they interacted in vivo) were scored computationally to determine which interactions were conserved in at least two species. The program was trained based on known conserved protein complexes and was able to successfully identify other known protein complexes as well as novel complexes.

key takeaways: The authors laid out their method and showed that they were able to successfully train their program to identify conserved protein complexes across five evolutionarily disparate species.

Figure 2: In this figure, the authors use their computational method to analyze protein co-complex interactions across the different taxa studied. They first show how they have expanded on their previous work in the current study by including new taxa and more human proteins (panel A). They also validate their method by demonstrating that they are able to use their co-fractionation experiments plus external evidence to more accurately identify co-fractionated proteins compared to either method alone (panel B). They measure accuracy based on precision (the fraction of results that were relevant) and recall (the fraction of relevant instances retrieved). They also show the success of their method by confirming that complexes with high co-elution scores were predicted to interact based on their algorithm.

The authors further validated their method by showing they were able to successfully detect distinct subunits of a known protein across three species (panel C). The protein they chose was the 30S subunit of the proteasome, which is a complex shaped like a tube with a lid on top. Co-fractionation data, a correlation matrix, and a hierarchical clustering dendrogram were all consistent with the known shape of the proteasome. Finally, the authors broke down the conserved co-complex interactions discovered to state whether they had previously been identified in other databases. They found that the majority of the PPIs they found were in fact novel.

key takeways: The authors’ method is valid and can be applied to find many novel PPIs.

Figure 3: In this figure, the authors looked at the conservation of the 981 multiprotein complexes identified by their method, as well as the origins of the subunits of these complexes. They identified the subunits as being either “old,” meaning the complexes had component ages of over one billion years of evolutionary divergence, or “new,” meaning the complexes were metazoan-specific (or only found only in animals) and had diverged evolutionarily approximately 500 million years ago (panel A). Many complexes were made of mixed components, meaning they contained subunits that were both old and new. This figure also shows “zoomed-in” versions of six example complexes that were either completely old, completely new, or mixed (panel B).

Finally, the authors showed the prevalence of the putative complexes across phyla by looking at the conservation of the six complexes from panel B across phyla (panel C). They saw that all of the complexes tended to have orthologs in animals (Cnidarians, Protostomes, and Deuterostomes). On the other hand, fewer complexes containing new subunits had orthologs in fungi, protists, and plants, as seen by the dark blue areas (indicating lack of orthlogs) for complexes 3-6 (which contained new subunits) in phyla that were more evolutionarily distant from Metazoa.

key takeaways: The rise of Metazoa, or animals, led to the development of many new genes that encoded for protein subunits involved in multicellularity. However, “older” protein subunits and complexes are still conserved across phyla and are generally involved in core cellular processes.

Figure 4: Here, the authors attempted to verify the accuracy of their predictions via physically analyzing the complexes they identified computationally. They performed affinity purification mass spectrometry (AP/MS) to verify the interactions among a number of their predicted complexes. This method uses one protein in the complex as “bait” to capture the other proteins by allowing them to bind to the bait. The proteins are then identified via MS. The results of these experiments showed that the predicted interacting proteins were almost always detected by human AP/MS, because the bait protein (the header in each box of the inset) caught its predicted interacting proteins in the vast majority of AP/MS trials (panel A).

The authors then used data from other papers to validate their results. They saw that the complexes they predicted to be conserved overlapped with another AP/MS experiment (panel B), and also that the inferred MW of the human protein complexes corresponded to size-exclusion chromatography profiles from other studies both across a wide range of sizes (panel C) and for one specific complex whose subunits were co-eluted (panel D).

key takeaways: The authors confirmed the accuracy of their predictions based on the physical properties of sample protein complexes that they had identified using their method.

Figure 5: In this final figure, the authors functionally validated a number of the interactions that they had identified previously in the paper. They first focused on a novel complex, Commander, a large multi-subunit protein in which the function of many subunits was previously unknown. They found that knocking down expression of the COMMD2 and COMMD3 subunits of this protein led to impaired head and eye development (panel A) as well as impaired brain development and neural patterning (panel B) in tadpoles. Next, the authors focused on the metazoan-specific PPI between BUB3 and ZNF207, which are involved in the proper attachment of chromosomes to spindles during cell division. They found that RNAi knockdown of either the C. elegans ZNF207 ortholog alone or BUB3 and ZNF207 together led to increased lethality for developing embryos. This had previously been seen in humans but not in worms (panel C). The authors then looked at mixed protein complexes. They found that knocking down the spliceosomal component involved in recognition of the branch site of unprocessed mRNA led to slower growth and increased sensitivity to a splicing inhibitor (panel D).

Finally, the authors took a network approach to these functional analyses. They found that their fractionation data were significantly enriched for proteins that interacted at sequential steps in signaling pathways (panel E), which makes sense because these proteins would have to interact during signaling. Consecutive interactions were also seen for widely conserved biosynthetic pathways such as purine synthesis (panel G), which involves metabolic channeling (when one enzyme passes its metabolic product directly to the active site of the next enzyme), as seen in panel F.

key takeaways: The authors’ method can be used to identify functionally relevant PPIs as well as important network interactions. The conservation of these PPIs can also be analyzed.

Reference:
Wan C, et al. 2015. Panorama of ancient metazoan macromolecular complexes. Nature. 525:339-344. Available from Macmillan Publishers.

My Page
Genomics Page
Biology Page

Email Questions or Comments: phparrish@davidson.edu