Human Protein Atlas of redox systems - what can be learnt?

Background: High-throughput screening projects are popular approaches to yield a vast amount of information amenable for database mining and "hypothesis generation". The keys to success for these approaches depend upon the quality of primary data, choice of algorithms for data analyses, solidity in data annotations and the general usefulness of the results. A large initiative aimed at mapping the expression of all human proteins is the Human Protein Atlas (www.proteinatlas.org), encompassing immunohistochemical analyses of human tissues utilizing antibodies raised against a large number of human proteins. Here, we wished to probe what could be learnt from this atlas using a manual in-depth analysis of the results regarding the expression of key proteins in the human glutathione and thioredoxin systems.

Methods: The freely available on-line data of immunohistochemical analyses for selected human redox proteins within the Human Protein Atlas were here analyzed, provided that reasonably solid data existed for the antibodies that were employed. This included tissue expression data for thioredoxin 1 (Trx1), Trx2, thioredoxin reductase 1 (TrxR1), TrxR2, glutathione reductase (GR), glucose 6-phosphate dehydrogenase (G6PD), γ-glutamyl cysteinyl synthase (gGCS) and the six peroxiredoxins Prx1 to Prx6. The data were further complemented with a screen using a polyclonal peptide antibody raised against the unique glutaredoxin domain of TXNRD1_v3 ("v3"). The results from fifteen major tissues and organs are presented (lung, kidney, liver, lymph node, testis, prostate, ovary, breast, pancreas, cerebellum, hippocampus, cerebral cortex, skin, skeletal muscle and heart muscle) and discussed considering earlier findings described in the literature.

Results: Staining patterns proved to be highly variable and often unexpected both in terms of tissues analyzed and the individual target proteins. Among the analyzed tissues, only macrophages of the lung, tubular cells of the kidney, lymphoid cells of lymph nodes, Leydig cells in the testis, glandular cells of the prostate and exocrine glandular cells of the pancreas, showed positive staining with all of the fourteen antibodies that were analyzed. Among these antibodies, those against Trx1, TrxR2 and G6PD showed the most restricted staining across different tissues, while others including the antibodies against Trx2, TrxR1, GR, Prx3, Prx4 and Prx6 gave strong staining in most tissues. Staining for v3 was strong in many cells and tissues, which was unexpected considering previous results mapping transcripts for this protein. No obvious co-variation in staining across tissues could be noted when comparing any two of the analyzed antibodies. Staining for G6PD was weak in most tissues, except for cells of the seminiferous ducts in testis and follicular cells of the ovary, where G6PD staining was strong.

Conclusions: Results from high-throughput screening projects such as the Human Protein Atlas must be taken with caution and need to be duly confirmed by thorough in-depth follow-up studies. The varying staining intensities comparing tissues as seen here for most of the analyzed antibodies nonetheless suggest that the overall profile of the human redox systems may vary significantly between different cell types and between different tissues.

General significance: The Human Protein Atlas data suggest that the individual proteins of the human thioredoxin and glutathione systems may be strikingly tissue- and cell type-specific in terms of expression levels, but we also conclude that these type of high-throughput results should be taken with significant caution and must be duly verified using subsequent focused and detailed hypothesis-guided follow-up studies. This article is part of a Special Issue entitled Human and Murine Redox Protein Atlases.