1
Science for Life Laboratory, College of Biotechnology,
KTH ‐ Royal Institute of Know-how,
Stockholm,
SE,
171 21,
Sweden,
2
Division of Immunology, Genetics and Pathology, Science for Life Laboratory, Rudbeck Laboratory,
Uppsala College,
Uppsala,
SE,
751 85,
Sweden,
Summary
Significance of Article
The article summarizes current updates and present standing of the Human Protein Atlas, www.proteinatlas.org, which is the most important and most complete database for spatial distribution of proteins in human tissues and cells. An outline of the publicly accessible database is offered, and its features and potential implications to be used in addition to the longer term path of spatial proteomics are mentioned.
Introduction
Proteins are the important constructing blocks of life, and resolving the spatial distribution of all human proteins on an organ, tissue, mobile, and sub‐mobile stage will drastically improve our understanding of human biology in well being and illness. Ever for the reason that completion of the human genome sequence, the last word objective has been to know the dynamic expression of the roughly 20,000 protein‐coding genes and to generate a map of the human proteome. Latest efforts embody the Human Proteome Map1 and the Proteomics DB2 primarily based on mass spectrometry of human tissues in addition to the initiative from the HUPO Human Proteome Undertaking (HPP), whose extra stringent pointers resulted in a extra correct map.3 A part of the HPP initiative is the Human Protein Atlas (HPA) mission, specializing in antibody‐primarily based proteomics and built-in omics.
An “atlas” is outlined as a set of maps or charts that provides a complete view on a sure topic. Beneath this premise, the objective of the publicly accessible HPA is to disclose the spatial distribution and expression of each human protein in several human tissues, most cancers varieties, and cell strains. This strategy permits single proteins and lists of proteins belonging to buildings akin to organs and organelles, or categorizing proteins primarily based on expression stage and tissue distribution, for instance, housekeeping proteins and tissue elevated proteins. A number of current achievements are a primary draft of a tissue‐primarily based atlas,4 a sub‐mobile atlas,5 and a pathology atlas.6
The HPA was initiated in 2003, and launched a primary model of the general public database www.proteinatlas.org in 2005, containing protein expression knowledge primarily based on roughly 700 antibodies.7 Since then, every new launch has included each extra knowledge and new web site functionalities, and main milestones encompass a gene‐centric database with info on all human genes predicted by Ensembl8 and addition of transcriptomics knowledge primarily based on excessive‐throughput mRNA sequencing.9 Each in‐home generated antibodies and business antibodies from completely different suppliers are used for immunohistochemistry (IHC) and immunofluorescence (IF). Model 17 comprises >25,000 antibodies which have handed rigorous high quality assessments for antigen specificity and validation, resulting in a set of greater than 10 million IHC pictures and 82,000 excessive‐decision IF pictures. Thereby, greater than 86% of the present 19,628 human protein‐coding genes in line with Ensembl model 83.3810 are already focused by a minimum of one antibody. Model 17 of the HPA is split into three sub‐atlases (Fig. (Fig.1):1): the Tissue Atlas describing expression and localization of proteins throughout 40 non‐diseased human organs utilizing RNA‐Seq and IHC on tissue microarrays (TMAs); the Pathology Atlas, containing RNA and protein expression knowledge for the 17 main varieties of human most cancers; and the Cell Atlas describing the sub‐mobile places of proteins to organelles with IF pictures in 22 cell strains and cell line‐particular gene expression throughout 56 completely different cell strains. The completely different sub‐atlases are interconnected and complement one another. This allows the person to discover a protein’s tissue and organ distribution, sub‐mobile localization, and relation to most cancers by toggling between the completely different sub‐atlases. The HPA supplies an essential useful resource for each primary and medical analysis, and within the current article, the completely different elements and features of the publicly accessible HPA webpage and first knowledge are offered and mentioned.
Antibody Validation
The experimentally decided protein places within the HPA are solely nearly as good as its fundamental reagent, the antibodies. Antibodies require excessive sensitivity and specificity to realize dependable knowledge, thus offering the perfect estimate of protein expression throughout tissues and cells. Consequently, antibody validation is a vital a part of the HPA. All antibodies produced inside the HPA mission must go high quality assurance steps earlier than being utilized in IHC and IF.11 First, plasmid inserts are sequenced to guarantee that the proper protein epitope signature tag (PrEST) sequence is cloned. Second, the scale of the ensuing recombinant protein (together with the particular PrEST) is thereafter analyzed utilizing mass spectrometry to guarantee that the proper antigen has been produced and purified. Third, to regulate for cross‐reactivity, affinity purified antibodies are examined for sensitivity and specificity on protein arrays consisting of glass slides with noticed PrEST fragments. HPA antibodies that meet these three standards must go a minimum of one extra assay earlier than they’re revealed on the Atlas. All of them in addition to commercially accessible antibodies are examined by Western blot evaluation of protein lysates from a restricted variety of tissues and cell strains. Photographs generated utilizing IHC and IF are critically evaluated and in contrast with accessible experimental gene/protein characterization knowledge. Antibodies that go these commonplace validation strategies are subsequently formally validated primarily based on the suggestions of the Worldwide Working Group for Antibody Validation committee,12 suggesting completely different “pillars” as commonplace for antibody validation. These pillars encompass genetic strategies (e.g., siRNA knockdown), impartial antibodies focusing on completely different epitopes of the identical antigen, orthogonal methods evaluating differentially expressed proteins (e.g., tissues with high and low expression) utilizing an antibody‐impartial technique, or expression of a fluorescent protein‐tagged protein. The strategies are described intimately on the HPA webpage (www.proteinatlas.org/about/antibody+validation) and associated assays could be discovered on the respective gene web page. Based mostly on how the antibody performs in several validation assays, all annotations are scored for his or her reliability at a 4‐tiered scale: “validated”, “supported”, “approved”, and “uncertain”. The present counts for the 4 classes primarily based on the 16,990 genes with a minimum of one accessible antibody within the HPA are 1548 validated (9.1%), 6012 supported (35.8%), 6927 permitted (40.8%), and 779 unsure (14.7%) genes. For genes the place the reliability rating differs between the Tissue Atlas and Cell Atlas, the rating with highest reliability is taken into account within the counts. Within the classes “approved” and “uncertain”, the variety of false annotation or off‐goal binding is of course greater, however the impact on world proteomic analyses is small.5 However, the person can filter for the reliability within the search area and obtain knowledge solely from genes with a sure reliability rating. The variety of genes with validated and supported antibodies will improve in upcoming releases of the HPA.
The right way to Get Began – “protein atlas”
In all three sub‐atlases, every gene has its personal abstract web page, which could be accessed in two alternative ways (Fig. (Fig.2).2). Essentially the most simple manner is the search perform [Fig. [Fig.2(A)],2(A)], which can be utilized without spending a dime textual content searches akin to gene identify, gene synonyms, gene descriptions or exterior gene and protein identifiers (UniProt, Ensembl, NCBI Entrez Gene) in addition to for searches primarily based on protein lessons, Gene Ontology identifiers and descriptions, antibody identifiers and picture annotations. For extra complicated queries, the “Fields” perform permits a selected seek for a listing of genes that match chosen traits. For instance, the search could be for a sure protein class, akin to, enzymes and receptors, predicted secreted proteins, or potential drug targets; or the search could be inside the major knowledge generated within the HPA about protein expression, or antibody validation in several assays. It’s not solely potential to incorporate (or exclude) proteins localized to a sure tissue or organelle, however to refine the search by combining a number of standards akin to including cell cycle dependent sub‐mobile expression and RNA expression. A search carried out through free textual content or “Fields” generates a gene‐centric listing of outcomes, which could be organized in a complete method depending on info of curiosity through the use of the “Show/hide columns” perform. A gene web page is accessed by clicking on a gene of curiosity, and the completely different sub‐atlases are reached by means of the corresponding thumbnail pictures [Fig. 2(B)].
The second solution to get to a gene web page is thru touchdown pages [Fig. [Fig.2(C)],2(C)], that are interactive data chapters discussing the proteome of a single compartment, akin to a single tissue or organelle. They include a short description of the tissue/organelle, summarize the HPA knowledge, and current instance pictures of the completely different cell varieties and morphologies. As well as, community plots present how the tissue/organelle is related with different tissues/organelles by proteins with an identical expression or multilocalization. A key function of the touchdown pages is their interactivity. Each picture, quantity, or plot is a clickable and leads both on to a listing of genes, a gene abstract web page, or a selected tissue or cell picture. Extra info on the touchdown pages within the completely different sub‐atlases is described under.
Tissue Atlas
The most important Tissue Atlas launch in 2014 included addition of RNA‐Seq knowledge, with every gene web page on the Tissue Atlas containing a complete abstract of expression each on the mRNA and protein stage.4 The protein expression knowledge, at present masking 15,297 (78%) of the protein‐coding genes, is derived from antibody‐primarily based protein profiling utilizing IHC on TMAs. Altogether 76 completely different cell varieties, equivalent to 44 non‐diseased human tissue varieties masking all main elements of the human physique, have been analyzed manually and the info is offered as histology‐primarily based annotation of protein expression ranges. Along with the usual setup, prolonged tissue profiling is carried out for chosen proteins, to offer a extra full overview of the place the protein is expressed. Prolonged tissue samples embody mouse mind, human lactating breast, eye, and extra samples of adrenal gland, pores and skin and mind. The present model comprises 3452 such pictures, and upcoming variations will embody each extra genes and extra varieties of organs with prolonged tissue profiling.
Determine Figure33 summarizes the format of a Tissue Atlas gene web page, exemplified by the gene CCNB1. On high of the gene web page [Fig. [Fig.3(A)],3(A)], three containers are discovered. “General information” summarizes gene info from Ensembl, protein class, predicted localization and variety of transcripts. By clicking on “Show more”, Entrez info and extra exterior hyperlinks to accessible gene identifiers are offered. Every heading with an “i”‐signal is clickable with a brief description of the content material. “HPA information” supplies a abstract of RNA tissue class primarily based on each internally generated RNA‐Seq knowledge (HPA), in addition to two exterior RNA expression datasets; RNA‐Seq knowledge from the Genotype‐Tissue Expression (GTEx) consortium13 and CAGE knowledge from the FANTOM5 consortium.14 The RNA tissue classes group all human protein‐coding genes primarily based on sample of expression, together with expressed in all, tissue enriched, group enriched, tissue enhanced, blended or not detected, as described beforehand in.9 RNA tissue classes are calculated individually for the three completely different RNA expression datasets, together with 37 tissues from HPA, 31 tissues from GTEx, and 35 tissues from FANTOM5, altogether masking 40 of the 44 tissues analyzed with IHC. Different info offered within the “HPA information” field consists of “Protein evidence” generated from a number of impartial sources. “Protein expression” is a brief abstract of the general protein expression profile in non‐diseased tissues, together with sub‐mobile localization and tissue distribution. Within the “Data reliability” field, the “Data reliability description” summarizes the data‐primarily based interpretation of the first knowledge, whereas “Reliability score” reveals a 4‐tiered reliability rating (see extra beneath Antibody validation part) and ID:s of the antibodies used within the assay. Clicking on “Show more” leads into the “Antibody validation” web page with detailed info on all antibody validation assays.
Beneath the three info containers, the “RNA and protein expression summary” is proven [Fig. [Fig.3(B)],3(B)], offering an outline of information generated within the HPA mission. The analyzed tissues are divided into 13 completely different teams in line with widespread useful options, and every group is clickable for entry to lists of included tissues. On the fitting panel of the “RNA and protein expression summary”, pictures of chosen tissues give a visible abstract of the protein expression. Beneath are separate panels displaying tissue particular expression in all analyzed tissues each on the protein stage (“Protein expression overview”) and the RNA stage (“RNA expression overview”) within the three completely different RNA expression datasets [Fig. [Fig.3(C)].3(C)]. Clicking on a tissue identify or bar supplies entry to the detailed knowledge web page.
The detailed knowledge web page (Fig. (Fig.4)4) is exclusive for every analyzed tissue and reveals pictures of the stained tissue samples, along with expression stage of the analyzed cell varieties. As right here exemplified by testis, three pictures every for the three completely different antibodies used within the assay are displayed, with protein expression summarized as extremely expressed in a subset of cells in seminiferous ducts and never detected in Leydig cells [Fig. [Fig.4(A)].4(A)]. All pictures are clickable for an enlarged excessive‐decision view, permitting for visible examination of the protein expression within the context of neighboring cells [Fig. [Fig.4(B)].4(B)]. Beneath, particulars on the RNA expression knowledge is proven [Fig. [Fig.4(C)],4(C)], with knowledge on expression in every of the analyzed people of the three completely different RNA expression datasets (HPA, GTEx and FANTOM5). For the samples analyzed within the HPA dataset, a hematoxylin and eosin stained picture from a consecutive part of the tissue materials used for RNA‐Seq is offered, together with estimated fractions of the cell varieties current within the pattern. This provides the person a risk to guage and additional perceive the RNA expression knowledge, which relies on a combination of various cell varieties, and examine the knowledge with cell‐sort particular protein expression profiles.
Quite a few complete touchdown pages on the HPA (www.proteinatlas.org/humanproteome) describe the proteome and transcriptome of every organ, in addition to sub‐proteomes equivalent to specific useful teams of genes, as summarized previously4, 15 [Fig. [Fig.2(C)].2(C)]. The tissue and organ proteome touchdown pages, such because the lung‐particular proteome,16 the liver‐particular proteome,17 the testis‐particular proteome,18 and many others. embody catalogs of proteins expressed in a tissue‐restricted method, primarily based on HPA RNA‐Seq knowledge. Such proteins are believed to play an essential function within the organ physiology and supply the premise for organ‐particular analysis in well being and illness. Every touchdown web page comprises full lists of expression of all genes in a sure organ, clickable with direct hyperlinks to go looking outcomes or Tissue Atlas gene abstract pages, the place proteins of curiosity could be explored additional. Community plots present group enriched genes, highlighting genes which are concurrently elevated in a gaggle of two–7 tissues, in comparison with all different analyzed tissues. The plots assist find widespread options between completely different organs, and additional elucidating the perform of group enriched genes.
Different touchdown pages discovered within the Tissue Atlas are the sub‐proteomes that summarize sure useful teams of genes. Such proteomes embody “the druggable proteome”, “the secretome and membrane proteome”, “the cancer proteome”, “the regulatory proteome”, and “the isoform proteome”, as described beforehand in Uhlén et al.4 All sub‐proteome touchdown pages summarize basic data about every proteome, present immunohistochemical examples and include quite a few clickable lists with entry to Tissue Atlas major knowledge.
Pathology Atlas
The earlier Most cancers Atlas contained protein expression knowledge for the 20 most typical varieties of most cancers stained with IHC on TMAs utilizing the identical workflow as for the Tissue Atlas. In model 17 of the HPA, the Most cancers Atlas modified its identify to Pathology Atlas along with a significant re‐design and launch of latest knowledge, displaying the affiliation of all human genes with medical final result.6 Within the Pathology Atlas, a methods stage strategy was used to research the human genome with respect to medical final result primarily based on genome‐huge expression knowledge from the Most cancers Genome Atlas.19 RNA‐Seq knowledge and medical metadata from 8,000 particular person sufferers equivalent to 17 of the 20 main most cancers varieties included within the HPA have been used for figuring out the correlation between RNA expression ranges and general survival time for every gene in every most cancers sort. Greater than 500,000 Kaplan‐Meier plots enable for unbiased identification of prognostic genes. By way of search fields and Pathology Atlas touchdown pages, lists of prognostic genes with a excessive significance (P < 0.001) are highlighted. The prognostic genes are additional divided into favorable genes, the place excessive RNA expression correlates with longer survival time, and unfavorable genes, the place excessive RNA expression correlates with shorter survival instances. The RNA‐Seq knowledge are additionally used for categorization of all genes in the identical method as within the Tissue Atlas, permitting for identification of genes elevated in a sure most cancers sort in comparison with different cancers. An outline of a gene abstract web page within the Pathology Atlas is proven in Determine Figure5,5, exemplified by CCNB1. For genes the place a major prognostic affiliation is discovered, Kaplan‐Meier plots for the most cancers varieties the place the gene is prognostic are proven within the “Prognostic summary” part [Fig. [Fig.5(A)].5(A)]. Beneath, RNA expression ranges throughout the 17 most cancers varieties [Fig. [Fig.5(B)]5(B)] are summarized. Within the “Protein expression” part, examples of IHC stained most cancers tissues are proven, and a abstract of protein expression ranges throughout completely different most cancers varieties analyzed with IHC is offered [Fig. [Fig.5(C)].5(C)]. The IHC evaluation of most cancers tissues is carried out on as much as 12 people every from the 17 most cancers varieties analyzed with RNA‐Seq, in addition to three extra most cancers varieties; nonetheless, the person samples aren't linked between the RNA‐Seq and IHC analyses. Each most cancers tissue has an in depth knowledge web page offering entry to survival evaluation knowledge and RNA expression ranges for every affected person [Fig. [Fig.5(D)],5(D)], in addition to clickable excessive‐decision IHC pictures of most cancers tissues, as much as 12 people per most cancers sort [Fig. [Fig.55(E)]. Pathology Atlas touchdown pages are offered for every most cancers sort, offering a complete overview of the genomic and proteomic panorama of every sort of most cancers, permitting entry to most cancers tissue elevated genes and prognostic genes. "protein atlas"