Multiclassifier combinatorial proteomics of organelle shadows at the example of mitochondria in chromatin data

Abstract

Subcellular localization is an important aspect of protein function, but the protein composition of many intracellular compartments is poorly characterized. For example, many nuclear bodies are challenging to isolate biochemically and thus remain inaccessible to proteomics. Here, we explore covariation in proteomics data as an alternative route to subcellular proteomes. Rather than targeting a structure of interest biochemically, we target it by machine learning. This becomes possible by taking data obtained for one organelle and searching it for traces of another organelle. As an extreme example and proof‐of‐concept we predict mitochondrial proteins based on their covariation in published interphase chromatin data. We detect about ⅓ of the known mitochondrial proteins in our chromatin data, presumably most as contaminants. However, these proteins are not present at random. We show covariation of mitochondrial proteins in chromatin proteomics data. We then exploit this covariation by multiclassifier combinatorial proteomics to define a list of mitochondrial proteins. This list agrees well with different databases on mitochondrial composition. This benchmark test raises the possibility that, in principle, covariation proteomics may also be applicable to structures for which no biochemical isolation procedures are available.

Compositional Dynamics: Defining the Fuzzy Cell

“…This led to the observation that function at a subcellular location can also be inferred from proteomics data alone. This follows a two-step procedure: first, proteins are quantified across multiple biochemical isolations of a cellular structure, obtained from differently perturbed cells as starting material 3, 8. Second, one determines the covariation of all identified proteins with known functional components of that organelle (Figure 1).…”

“…The ‘behavior similarity’ or covariation can be measured using multi-classifier combinatorial proteomics (MCCP) [1], which is based on another machine-learning approach, random forests. So far, both chromatin components 1, 3 and mitochondrial proteins [8] can be determined on the basis of their covariation, suggesting this could be a general method to determine functional organelle composition and an alternative to approaches based on co-fractionation. Indeed, covariation was better suited to distinguish functional from non-relevant chromatin-bound proteins than classical, purification-based approaches [3].…”

“…Indeed, covariation was better suited to distinguish functional from non-relevant chromatin-bound proteins than classical, purification-based approaches [3]. Protein covariation can also inform on organelle composition for organelles that contaminate the biochemical purification of another organelle [8]. In principle, the more different biological conditions that are tested for the composition of an organelle, the better one can capture its constitutive, functional components.…”

Quantifying compartment‐associated variations of protein abundance in proteomics data

“…Additionally, it is known that different tissues show different levels of respiratory activity and variable amounts of mitochondria (Kirby et al , 2007; Fernández‐Vizarra et al , 2011) and that the number and composition of organelles can be affected, for example, by the aging process (Cellerino & Ori, 2017). Covariation of protein abundance across different conditions can be also exploited, and it can contribute to functional proteomics (Kustatscher et al , 2016). Currently, there is a lack of systematic approaches able to detect and deal with differences in cellular organization that might influence the outcome of proteomics data analysis from unfractionated samples.…”

The human proteome co-regulation map reveals functional relationships between proteins

“…Proof-of-principle studies by us and others have shown that protein covariation can be used to infer, for example, the composition of protein complexes and organelles [34][35][36][37][38][39][40][41][42] . However, these studies have focussed on relatively small sets of proteins or biological conditions, or used samples tailored to the analysis of specific cellular structures.…”