Correlations between RNA and protein expression profiles in 23 human cell lines

Abstract

Background: The Central Dogma of biology holds, in famously simplified terms, that DNA makes RNA makes proteins, but there is considerable uncertainty regarding the general, genome-wide correlation between levels of RNA and corresponding proteins. Therefore, to assess degrees of this correlation we compared the RNA profiles (determined using both cDNA- and oligo-based microarrays) and protein profiles (determined immunohistochemically in tissue microarrays) of 1066 gene products in 23 human cell lines.

Results: A high mean correlation coefficient (0.52) was obtained from the pairwise comparison of RNA levels determined by the two platforms. Significant correlations, with correlation coefficients exceeding 0.445, between protein and RNA levels were also obtained for a third of the specific gene products. However, the correlation coefficients between levels of RNA and protein products of specific genes varied widely, and the mean correlations between the protein and corresponding RNA levels determined using the cDNA- and oligo-based microarrays were 0.25 and 0.20, respectively.

Conclusion: Significant correlations were found in one third of the examined RNA species and corresponding proteins. These results suggest that RNA profiling might provide indirect support to antibodies' specificity, since whenever a evident correlation between the RNA and protein profiles exists, this can sustain that the antibodies used in the immunoassay recognized their cognate antigens.

Figure 1

Outline of the experimental procedure. RNA expression profiles were generated using cDNA and oligo microarrays, and protein expression levels were generated using immunohistochemical staining of cell microarrays with antibodies from the Human Protein Atlas initiative. The expression levels were measured in each assay in each of the 23 cell lines. For each of the 1066 gene products for which data were obtained from all three platforms, the Spearman correlation coefficients between the RNA _oligo-protein, RNA _cDNA-protein and RNA _oligo-RNA _cDNA datasets were calculated. The equation of the Spearman correlation calculation is shown in the Figure.

Figure 2

Four examples of different correlation coefficients: 0.743, 0.498, 0.240 and 0.00 (from the top left to the bottom right) between expression levels of RNA and protein gene products with corresponding Ensembl IDs (ENSG00000106415, ENSG00000175305, ENSG00000130522 and ENSG00000072041, respectively). The two lines indicate RNA expression levels measured in oligo microarray (blue) and protein expression analyses (red), across 23 cell lines. The values shown have been adjusted to a comparable scale, by adding the absolute value of the lowest RNA oligo value (which is always negative) to all RNA oligo values. All values (RNA oligo and protein) have then been divided by the highest value for the RNA oligo data. This gives measurements on a comparable (0 – 1) scale and the correlation coefficient remains the same as before the adjustment.

Figure 3

Histograms of all correlation coefficients for each gene product obtained from each of the three comparisons, and one showing those of randomly picked Ensembl ID pairs. Figures 3a – 3c show the RNA _oligo versus protein profiles, RNA _cDNA versus protein profiles, and the RNA _oligo versus RNA _cDNA profiles, which yielded mean correlation coefficients of 0.25, 0.20 and 0.52, respectively. The distributions of correlation coefficients between RNA values obtained using both RNA platforms and the protein values have Gaussian shapes, but with some bimodal characteristics, in which most of the data points are centered at the respective mean, but shoulders can be seen at correlation coefficients of ~0.5 – 0.7. For the RNA assay correlations the distributions follow a beta distribution. The randomly picked pairs have a mean value close to zero, indicating that there is no apparent bias in the data set.

Figure 4

Venn diagram showing numbers of highly correlating gene products identified in the comparison of data obtained from each permutation of platforms. The gene products that had correlations 0.445 in each comparison (RNA_oligo-protein, RNA_cDNA-protein and RNA_oligo-RNA_cDNA) were compared with those identified in the other assays. 169 gene products with such correlations were identified in all three comparisons, equivalent to 63% and 82.5% of those found in the RNA _oligo-protein and RNA_cDNA-protein comparisons, respectively. The numbers of gene products with correlation coefficients 0.445 in the RNA _oligo-protein, RNA_cDNA-protein and RNA_oligo-RNA _cDNA comparisons were 292, 238 and 678, respectively.

Figure 5

Dendrograms of hierarchical clusterings based on 1 – Pearson correlation coefficient metrics. The expression levels of 169 Ensembl gene IDs with correlation coefficients >0.445 for which data were available in all three comparisons across 23 cell lines measured in each assay were utilized to cluster the data into three individual clusters. The cell lines are colored depending on their origins; red, deep-red, grey, green and blue indicate cells of: lymphoid; myeloid; melanoma, glioma and sarcoma; carcinoma and neuronal origins, respectively.