RNA-Validation of Antibody Performance for Proteome Research

By: Linn Fagerberg1, Åsa Sivertsson1, Mathias Uhlen1, Fredrik Ponten2, and Anna Asplund2, Poster - The Human Protein Atlas
1Science for Life Laboratory, Royal Institute of Technology, Stockholm, Sweden
2Dept of Immunolog, Genetics, and Pathology, Science for Life Laboratory, Uppsala University, Uppsala, Sweden

Download the poster (5.4 Mb PDF)

Introduction

Immunohistochemistry (IHC) enables spatial mapping of protein expression, with simultaneous presentation of morphology and histology.

The Human Protein Atlas project works systematically towards completing a proteomic map of gene expression in a range of normal and cancer issues using affinity-purified polyclonal antibodies and IHC1. The project also generates protein expression data in 46 cell lines. In version 10.0 of the atlas, released in September 2012, IHC data for 70% for the human proteome will be included. www.proteinatlas.org

All antibodies are validated using protein array, WB, IHC, IF and when applicable APE-annotated protein expression of two or more antibodies. Here we present an additional strategy for quality assessment of antibody performance:
Transcript-based validation performed in a set of 17 cell lines

Material and Methods

Protein data
Cell lines in issue microarray format, as described in (2).
17/46 cell lines used included in this study
10184 antibodies stained for relative protein quantification.
Protein quantification using automated image analysis (TMAx) (3).
(see examples of TMAx overlays to the right)

RNA data
RNAsequencing using Illumina HiSeq 2000 system.
FPKM values used for comparison of RNA and protein levels.

Approval strategy

  • All antibodies yielding a correlation of ≥0.5 (pair-­wise Spearman)
  • All antibodies detected above threshold levels on both RNA and protein level in the same cell lines (congruent expression pattern)

 


Data

Figure 1. a) Pairwise plot of IHC and RNAseq quantitative data in 17 cell lines for GATA1, along with example IHC images of positive hematopoietic cell line K-562 and negative breast cancer cell lines MCF-7.

Pairwise plot of IHC and RNAseq quantitative data in 17 cell lines for GATA1

 


 

Figure 2. Comparison of performance in WB between RNA-approved and not approved antibodies. Performance scored as: 1-2 supportive; 3-5 uncertain; 6-8 non-supportive.

Comparison of performance in WB between RNA-approved and not approved antibodies

 


 

Results and Discussion

42% (4290) of antibodies approved

  • 18% correlation ≥0.5
  • 24% congruent expression pattern

Although the methods (IHC and RNAseq) display vastly different levels of sensitivity and resolution, and although mechanisms controlling protein levels are likely to be complex, data on RNA abundance presents a valuable complement to the validation scheme. Particularly for not yet characterized proteins.

Alternative strategies, as for example comparing the staining patterns generated with two antibodies targeting non–overlapping epitopes on the same proteins, will together with RNA-validation constitute important validation strategies in the HPA effort to present valid mapping of the human proteome.

 

 

 

 

References

  1. Uhlen et al, Moll Cell Prot 4:1920-­1932, 2005.
  2. Andersson et al, JHC 54(12):1413-­1423, 2006.
  3. Stromberg et al, Proteomics 2007, 7, 2142-­2150.