Analysis of human MDM4 variants in papillary thyroid carcinomas reveals new potential markers of cancer properties

Abstract

A wild-type (wt) p53 gene characterizes thyroid tumors, except for the rare anaplastic histotype. Because p53 inactivation is a prerequisite for tumor development, alterations of p53 regulators represent an alternative way to impair p53 function. Indeed, murine double minute 2 (MDM2), the main p53 negative regulator, is overexpressed in many tumor histotypes including those of the thyroid. A new p53 regulator, MDM4 (a.k.a. MDMX or HDMX) an analog of MDM2, represents a new oncogene although its impact on tumor properties remains largely unexplored. We estimated levels of MDM2, MDM4, and its variants, MDM4-S (originally HDMX-S) and MDM4-211 (originally HDMX211), in a group of 57 papillary thyroid carcinomas (PTC), characterized by wt tumor protein 53, in comparison to matched contra-lateral lobe normal tissue. Further, we evaluated the association between expression levels of these genes and the histopathological features of tumors. Quantitative real-time polymerase chain reaction revealed a highly significant downregulation of MDM4 mRNA in tumor tissue compared to control tissue (P<0.0001), a finding confirmed by western blot on a subset of 20 tissue pairs. Moreover, the tumor-to-normal ratio of MDM4 levels for each individual was significantly lower in late tumor stages, suggesting a specific downregulation of MDM4 expression with tumor progression. In comparison, MDM2 messenger RNA (mRNA) and protein levels were frequently upregulated with no correlation with MDM4 levels. Lastly, we frequently detected overexpression of MDM4-S mRNA and presence of the aberrant form, MDM4-211 in this tumor group. These findings indicate that MDM4 alterations are a frequent event in PTC. It is worthy to note that the significant downregulation of full-length MDM4 in PTC reveals a novel status of this factor in human cancer that counsels careful evaluation of its role in human tumorigenesis and of its potential as therapeutic target.

Fig. 1

a–d Frequency histograms showing the distribution of the mRNA levels of the indicated genes in the indicated sample groups. Relative units of target mRNA are referred to a sample called calibrator, chosen to represent 1× expression of the target gene. Each sample mRNA was normalized relative to the β-actin mRNA. The superimposed line shows the normal distribution curve. For each distribution, the mean (M) ± standard deviation (SD) is shown. e, f Comparison of MDM4 and MDM2 levels, respectively, between control and tumor samples. Each plot shows graphically the central location and scatter–dispersion of the values of each group: the line series shows parametric statistics (mean and confidence interval of mean) while the notched box and whiskers show nonparametric statistics (median, confidence interval of median, and interquartile range). Crosses and circles indicate possible outliers, between 1.5 and 3 interquartile range and over 3 interquartile range, respectively. p value was calculated according to Wilcoxon signed-ranks test

Fig. 2

Receive operator characteristics curves for MDM4 levels between normal and tumor samples, assessing variable MDM4 levels as diagnostic test. The plot shows sensitivity of the test against specificity. p value was 0.0001. For each decision threshold, the percentage of tumor cases correctly identified as such (true positives) against the percentage of normal cases incorrectly diagnosed as tumor (false positives) is shown

Fig. 3

Western blot of WCE derived from 20 couples of tumor (T) and matching control (N). The values derive from the ratio of densitometric value of MDM4–actin, both at the protein and mRNA levels. Asterisks mark tumor samples expressing MDM4-211 mRNA; section signs mark tumor samples showing increased levels of MDM2 protein with no increased levels of the corresponding mRNA. Arrows mark a lower weight protein of about 58 kDa, recognized by different αMDM4 antibodies

Fig. 4

a, b Frequency histograms showing the distribution of the T-to-N ratios derived from tumor–CTRL levels of the indicated genes. The superimposed line shows the normal distribution curve. For each distribution, the mean ± standard deviation is shown

Fig. 5

a Vertical box-whisker plots showing MDM4 levels between nonmultifocal (no) and multifocal (yes) tumors. Each plot shows graphically the central location and scatter–dispersion of the values of each group: the line series shows parametric statistics (mean and confidence interval of mean) while the notched box and whiskers show nonparametric statistics (median, confidence interval of median, and interquartile range). Crosses and circles indicate possible outliers, between 1.5 and 3 interquartile range and over 3 interquartile range, respectively. p value was calculated according to Spearman test. b Vertical box-whisker plots showing tumor–CTRL MDM4 mRNA levels in stage 1 compared to other stages of tumors. Each plot shows graphically the central location and scatter–dispersion of the values of each group: the line series shows parametric statistics (mean and confidence interval of mean) while the notched box and whiskers show nonparametric statistics (median, confidence interval of median, and interquartile range). Crosses and circles indicate possible outliers, between 1.5 and 3 interquartile range and over 3 interquartile range, respectively. p value was calculated according to Spearman test