Differentially expressed proteins in malignant and benign adrenocortical tumors

Abstract

We have compared the microsomal protein composition of eight malignant and six benign adrenocortical tumors with proteomic methods. IGF2 had increased level in the malignant tumors, confirming previous microarray studies on the same material. Aldolase A, a glycolytic enzyme, also showed increased levels in the malignant tissue compared to the benign. Additionally, several proteins belonging to complex I in the mitochondrial respiration chain showed decreased levels in the malignant tissue. Taken together, this may indicate a shift in energy metabolism where glycolysis may be favored over tight coupling of glycolysis and mitochondrial respiration, a phenomenon known as the Warburg effect. One of the complex I proteins that showed decreased levels in the malignant tissue was GRIM-19. This protein has been suggested as a tumor suppressive protein by being a negative regulator of STAT3. In summary, an analysis of the microsomal proteome in adrenocortical tumors identifies groups of proteins as well as specific proteins differentially expressed in the benign and malignant forms. These proteins shed light on the biology behind malignancy and could delineate future drug targets.

Figure 1. Illustration of the experimental procedure.

The 14 samples (eight ACCs and six ACAs) were run in two separate 8-plex iTRAQ sets. An internal standard was included in each iTRAQ set. See Experimental procedures for details.

Figure 2. Enrichment analysis.

Panel A shows a pie chart depicting the number of transmembrane segments the 1902 identified and quantified proteins are predicted to contain (from ProteinCenter). Panel B shows the GOrilla results performed on the 1902 proteins, ranked by t-test p-value. Shown are enriched GO terms in “cellular component”.

Figure 3. Multivariate data analysis and outlier detection.

Panel A shows a PCA plot of the data. Sample 1151 was identified as an outlier. Panel B shows the underlying data that formed the PCA in panel A. Circled are the proteins causing the deviating behavior of sample 1151. These proteins were run through the Ingenuity Pathway Analysis program. In panel C the top canonical pathway is shown; acute phase pathway. A PCA without sample 1151 is shown in panel D; no more outliers were found.

Figure 4. Visualization of the data analysis workflow.

A) Both univariate (Student's t-test) and multivariate (OPLS) analyses were performed. B) The top canonical pathway identified by Ingenuity Pathway Analysis was mitochondrial dysfunction. Many proteins in this pathway were found to be downregulated in the malignant samples. C) Hierarchical clustering of the overlapping proteins (t-test and OPLS). Class 1: ACA, class 2: ACC.

Figure 5. Western blot and immunohistochemical analyses.

A–D) Expression levels of aldolase A and GRIM-19. To the left are the expression levels from the MS data and to the right are the western blot analyses. Ku70 and actin are loading controls. Western blot analyses were performed on tissue samples that showed the most significant differences in the iTRAQ experiments (two ACAs and two ACCs). E–F) Immunohistochemical analyses with anti-GRIM-19 (panels E and F). GRIM-19 staining in ACAs had a grain-like pattern, suggesting mitochondrial localization (panel E). In ACCs there was a more cytoplasmic staining (panel F).

Figure 6. Correlation between protein expression levels and tumor size.

Protein expression levels of the 26 proteins that overlapped in the t-test and OPLS analyses correlate with the size of the tumors. Two proteins have increased expression levels (light grey dots/lines), the rest have decreased expression levels (black dots/lines). Corresponding protein names can be found in Table S4.