Validation of housekeeping gene and impact on normalized gene expression in clear cell renal cell carcinoma: critical reassessment of YBX3/ZONAB/CSDA expression

Abstract

Background: YBX3/ZONAB/CSDA is an epithelial-specific transcription factor acting in the density-based switch between proliferation and differentiation. Our laboratory reported overexpression of YBX3 in clear cell renal cell arcinoma (ccRCC), as part of a wide study of YBX3 regulation in vitro and in vivo. The preliminary data was limited to 5 cases, of which only 3 could be compared to paired normal tissue, and beta-Actin was used as sole reference to normalize gene expression. We thus decided to re-evaluate YBX3 expression by real-time-PCR in a larger panel of ccRCC samples, and their paired healthy tissue, with special attention on experimental biases such as inter-individual variations, primer specificity, and reference gene for normalization.

Results: Gene expression was measured by RT-qPCR in 16 ccRCC samples, each compared to corresponding healthy tissue to minimize inter-individual variations. Eight potential housekeeping genes were evaluated for expression level and stability among the 16-paired samples. Among tested housekeeping genes, PPIA and RPS13, especially in combination, proved best suitable to normalize gene expression in ccRCC tissues as compared to classical reference genes such as beta-Actin, GAPDH, 18S or B2M. Using this pair as reference, YBX3 expression level among a collection of 16 ccRCC tumors was not significantly increased as compared to normal adjacent tissues. However, stratification according to Fuhrman grade disclosed higher YBX3 expression levels in low-grade tumors and lower in high-grade tumors. Immunoperoxidase confirmed homogeneous nuclear staining for YBX3 in low-grade but revealed nuclear heterogeneity in high-grade tumors.

Conclusions: This paper underlines that special attention to reference gene products in the design of real-time PCR analysis of tumoral tissue is crucial to avoid misleading conclusions. Furthermore, we found that global YBX3/ZONAB/CSDA mRNA expression level may be considered within a "signature" of RCC grading.

Figure 1

Comparison of expression level for the 8 indicated housekeeping genes (HKG) in the 16 ccRCC samples (grey boxes) and their paired normal tissue (open boxes). Values are cycle threshold (Ct) cross-points as defined in Material and Methods. Boxes are from lower to upper quartiles intersected by medians; whiskers are Min/Max values in the cohort of samples. GAPDH, 18S, B2M and RPL27 significantly differed between paired control versus tumoral groups (*p < 0.05 by Wilcoxon Test, see Table 3).

Figure 2

Effects of primer choice of reference HKG on the estimation of normalized YBX3 expression in ccRCC samples. Circles represent expression ratios distribution of YBX3 mRNA (YBX3-Ψ, filled circles; YBX3 + Ψ, open circles) in each of the 16 tumors compared to its normal paired tissue (set at 1, dot line) after normalization by beta-Actin (A) versus PPIA + RPS13 (B). Values are means of 5 independent assays (for the sake of clarity, standard deviations and statistical significance for each sample are not shown, but are provided in the histograms of Additional file 2). Note that pairs 1 to 3 were analyzed in the previous report [17]. Lines are medians with indicated values. After normalization by beta-Actin, YBX3 appears significantly overexpressed in tumors taken as a whole (A, **p < 0.01, *p < 0.05, Wilcoxon test) while it is not the case when normalized by PPIA and RPS13 (B, NS: not significant, Wilcoxon test).

Figure 3

Expression ratios of other cancer-related genes normalized to PPIA and RPS13 housekeeping genes. Boxes represent the lower and upper quartiles with medians in the 16 tumor samples compared to paired normal tissues (set at 1, red dot line); whiskers indicate the Tukey confidence intervals and black dots (•) are outlier values. VEGFa, c-Myc, CycD1 and megalin are significantly overexpressed in tumors (***p < 0.001, **p < 0.01, *p < 0.05, Wilcoxon Test) in comparison to PCNA and YBX3, which are globally not different from paired normal tissues. Beta-Actin expression is variable in tumor samples with a tendency to be down-regulated. (Please note the logarithmic Y axis due to strong c-Myc and Megalin overexpression).

Figure 4

Distribution of YBX3 expression ratios in tumor samples according to histological grading. (A) Circles are values for YBX3-Ψ expression ratios in the 15 Fuhrman-graded tumors compared to paired adjacent healthy tissue after normalization to geometric mean of PPIA + RPS13. Lines are medians for each group and statistical differences between subgroups are indicated as well (Mann Whitney test, *p < 0.05, **p < 0.01). The expression of YBX3 inversely correlates with tumor grade. (B) Representative images of YBX3 protein staining in tumor samples. Note (i) absence of nuclear labeling in stromal nuclei; (ii) overall decrease of intensity from grade #1 to #3; (iii) occasional cytoplasmic labeling in grade #2; and (iv) marked heterogeneity of nuclear labeling in some fields of grade #3 only. Scale bar; 50 μm.