Spatial partitioning of biological processes is a phenomenon fundamental to life that enables multiple processes to occur in parallel. An organelle is a sub module of the eukaryotic cell with a specialized function. The name "organelle" stems from the analogy between the role of organelles in the cells to the role of organs in the human body. The precise definition of organelles varies, and these sub modules are sometimes also referred to as compartments or structures of the cell. Often a distinction is made between membrane-bound and non-membrane bound organelles. The membrane-bound organelles, such as the nucleus and the Golgi apparatus, create a physical boundary thus separating the intra and extra-organelle space. In contrast, non-membrane bound organelles like the cytoskeleton and nucleoli provide a specialized surface or region. Membranous or not, this partitioning creates a specific environment at the site of the organelle, where the concentration of different molecules can be tailored to fit the purpose of the organelle.
At the cellular level, proteins function to catalyze, conduct and control most processes at specific times and locations. Subcellular localization of a protein helps define the protein function as different organelles offer distinct environments containing a variety of physiological conditions, and interaction partners. Consequently, mis-localizations of proteins have often been associated with cellular dysfunction and disease
(Kau TR et al, 2004;
Laurila K et al, 2009;
Park S et al, 2011). Knowledge of the spatial distribution of proteins at a subcellular level is thus essential for understanding protein function, interactions and cellular mechanisms. And studying the activity of how cells generate and maintain their spatial organization is central to understanding the mechanisms of the living cell.
Within the Cell Atlas, the subcellular localization of 12036 proteins have been mapped on a single-cell level to 32 subcellular structures and enabled the definition of 13 major organelle proteomes. The localization was performed in a panel of 22 human cell lines using transcriptomics data as a starting point. The analysis further reveals that approximately half of the proteins localize to multiple compartments and identifies many proteins with cell-to-cell variation in terms of protein abundance or spatial distribution. The expression pattern and spatial distribution of human proteins in all major cellular organelles can be explored in these interactive knowledge sections, including numerous catalogues of proteins with specific and similar patterns of expression, as well as examples of detailed images illustrating the subcellular spatial distribution patterns.
In the Cell Atlas, we employ an immunofluorescence (IF) based approach combined with confocal microscopy to enable high-resolution investigation of the spatial distribution of each protein
(Stadler C et al, 2013;
Barbe L et al, 2008;
Stadler C et al, 2010;
Fagerberg L et al, 2011). With the diffraction-limited resolution of about 200 nm, an immunofluorescence image from the Cell Atlas gives a detailed insight into the cellular organization. The spatial distribution of the protein is investigated using indirect IF in the U-2 OS cell line and up to two additional cell lines selected based on RNA-seq data. The protein of interest is visualized in green, while reference markers for microtubules (red), endoplasmic reticulum (yellow) and nucleus (blue) outline the cell. From small dots like nuclear bodies, to larger structures such as the nucleus, the distinct patterns in the images together with the reference markers make it possible to precisely determine the spatial distribution of a protein within the cell. This enables the assignment of the protein's location to one or more of the 32 structures and substructures currently annotated, as exemplified in Figure 1.
Figure 1. Example of confocal immunofluorescence images of different proteins (green) localized to each of the subcellular organelles and substructures currently annotated in the Cell Atlas in a representative set of cell lines. Microtubules are marked with an anti-tubulin antibody (red) and the nucleus is counterstained with DAPI (blue). The side of an image represents 64 μm. For more example images and details describing all the 32 patterns annotated in the Cell Atlas, see the Cell Dictionary.
Protein distribution in the human cell
Figure 2 shows the organelle distribution of all annotations for the 12036 proteins localized to at least one structure or substructure. The plot is sorted by meta-compartments: cytoplasm, nucleus, and secretory machinery, respectively. Most proteins are found in the nucleus, followed by the cytosol and vesicles, which consist of transport vesicles as well as small membrane-bound organelles like endosomes or peroxisomes. 51% (n=6190) of the proteins were detected at more than one location (multilocalizing proteins), and 15% (n=1862) displayed a (cell-to-cell variation) in expression level or spatial distribution. Explore the organelle proteomes of the human cell in detail here.
Figure 2. Bar plot showing the distribution of proteins detected in every organelle, structure and substructure annotated in the Cell Atlas.
Validation of antibodies and location data for the Cell Atlas
Recently, the quality and use of antibodies in research have been frequently debated
(Baker M. 2015). As antibody off-target binding can cause false positive results, we have made an effort in manually annotating all results regarding reliability of the staining. In the Cell Atlas a reliability score for every annotated location at a four-graded scale is provided: validated, supportive, uncertain and unreliable, as described in detail in the assay & annotation section. The validated locations are obtained through antibody validation according to one of the validation "pillars" proposed by an international working group
(Uhlen M et al, 2016): (i) genetic methods using siRNA silencing
(Stadler C et al, 2012) or CRISPR/Cas9 knock-out, (ii) expression of a fluorescent protein-tagged protein at endogenous levels
(Skogs M et al, 2016) or (iii) independent antibodies targeting different epitopes
(Stadler C et al, 2010). A supportive location is defined by agreement with external experimental data (UniProt database). An uncertain location score indicates that there is no external experimental information available to confirm the observed location. An unreliable location shows contradictory results compared to complementary information, such as literature or transcriptomics data. Also unreliable locations are shown, since it cannot be ruled out that the data is correct, and further experiments are needed to establish the reliability of the antibody staining. The distribution of reliability scores for the localized proteins is shown in Figure 3. Approximately 44% (n=5352) of the protein localizations provided are validated or supportive. Table 1 details the organelle distribution of all localized proteins and the distribution of reliability scores per organelle.
Figure 3. Pie chart showing level of reliability of the localized proteins, where each piece is the number of proteins with one type of score, out of the four reliability scores validated, supportive, uncertain and unreliable.
Table 1. Table showing the number of proteins localized to every organelle, structure, and substructure in the Cell Atlas, along with the distribution of reliability scores.
Stadler C et al, 2012. Systematic validation of antibody binding and protein subcellular localization using siRNA and confocal microscopy.J Proteomics.
PubMed: 22361696 DOI: 10.1016/j.jprot.2012.01.030
Stadler C et al, 2013. Immunofluorescence and fluorescent-protein tagging show high correlation for protein localization in mammalian cells.Nat Methods. 2013 Apr;10(4):315-23
PubMed: 23435261 DOI: 10.1038/nmeth.2377