Cancer Cell, Volume 12

Supplemental Data

Ultraconserved Regions Encoding ncRNAs Are

Altered in Human Leukemias and Carcinomas

George A. Calin, Chang-gong Liu, Manuela Ferracin, Terry Hyslop, Riccardo Spizzo, Cinzia Sevignani, Muller Fabbri, Amelia Cimmino, Eun Joo Lee, Sylwia E. Wojcik, Masayoshi Shimizu, Esmerina Tili, Simona Rossi, Cristian Taccioli, Flavia Pichiorri, Xiuping Liu, Simona Zupo, Vlad Herlea, Laura Gramantieri, Giovanni Lanza, Hansjuerg Alder, Laura Rassenti, Stefano Volinia, Thomas D. Schmittgen, Thomas J. Kipps, Massimo Negrini, and Carlo M. Croce

Table of Contents

I. Supplemental Experimental Procedures

II. Supplemental Figures

III. Supplemental Tables

IV. Supplemental References

I. Supplemental Experimental Procedures

Statistical Analyses for negative correlations between microarray expression of UCRs

and miRNAs.

The input data were constituted by a list of T-UCRs and by a list of miRNAs (the

"seeds") and the corresponding matrix of expression values. We calculated r, the Spearman

rank coefficient of correlation, a non-parametric measure of data trend correlation based on

rankings, for each pair of (miR, UCR) genes; namely, we evaluate the P-values of the

correlation tests and selected the genes whose correlation value is significant at a given value

of rejection. Evaluation of P-values was performed assuming that the correlation values are

distributed using Student's t cumulative distribution, with a number of degrees of freedom

corresponding to the number of samples in the microarray experiment. The P-values measure

the 'goodness' of the single correlations (among couples of genes), therefore, to understand if

the real correlation derives by chance or represents a biologically important information, we

choose the method of permutations, changing the order of the samples for each row (miR or

UCR) and calculating the correlations between pair of genes (miR, UCR) with different changed

samples orders. We repeated the samples permutation and computed correlations 100 times, in

this way, every real correlation has 100 random correlations to compare with. Using all (100 * n°

MIR * n° UCR) random correlations and real correlations, we recalculated P-values based on

random correlations ranking and position of the real correlations.

Given the high number of correlation tests performed, P-values were corrected for

multiple testing by using the false detection rate (FDR), as defined by (Benjamini and Hochberg,

1995). In this way, P-values control the number of false positive over the number of truly null

tests, while FDR controls the number of false positive over the number of significant tests.

Several ways of estimating this number have been proposed, and we adopted the solution

devised by Tom Nichols (see http://froi.sourceforge.net/documents/technical/matlab/FDR.html),

that rescales the P-value obtained on a single test multiplying it by a combination of indexes

related to the total number of tests performed. Correction was performed on a seed by seed

basis, meaning that the genes in the seeds list were considered independent tests. This

statistically validated tripe filtering allows the targeted extraction of a shortlist of candidate

genes, thus saving resources for the following costly and time-consuming genetic analysis.

To build a scatter plot between miR-24-1 and uc.160 expression values and between

miR-155 and uc.346+(A) expression values, respectively, we plotted a regression line by using

MatLab function ROBUSTFIT to explain hypotheses of negative correlation between these two

genes.

II. Supplemental Figures

Figure S1. UCR expression in human normal and malignant tissues by Northen blot. The

expression for uc.192(N) and uc.246(E) in normal mononuclear cells (MNC) and CLL samples is

presented. Normalization was done with U6 probe. The arrows on the left side show the

identified transcripts under the gel image there are the averages of UCR normalized expression

values in CLL and MNC samples from microarray experiments; p-values were from ANOVA

statistic.

Figure S2. UCR expression in human normal and malignant tissues by qRT-PCR. Relative

gene expression by qRT-PCR in CD5+/CD19+ positive lymphocytes and human chronic

lymphocytic leukemia (CLL) samples. UCR microarray values of CLL and CD5+ samples are

indicated under the graph; p-values were from ANOVA statistic.

Figure S3. T-UCR expression profile of 22 normal human tissues. Clustering of tissues and

UCRs revealed a distinct pattern of UCRs expression in normal human tissues. Samples are

shown in columns, T-UCRs in rows. A green colored gene is down-regulated compared to its

median expression in all samples, red is up-regulated and yellow means no variation.

Figure S4. T-UCR expression profile of 173 cancers and corresponding normal tissues.

Samples from leukemia and normal blood cells are separated from cancer and tissue of

epithelial origin. Samples are shown in columns, T-UCRs in rows. A green colored gene is

down-regulated compared to its median expression in all samples, red is up-regulated and

yellow means no variation.

Figure S5. The expression of uc.73A(P) gene in various colon cancer cell lines by

quantitative RT-PCR. The expression in normal colon represents the median value of 4

different samples. For normalization we used beta-actin.

Figure S6. The uc.73A(P) inhibition by siRNA1 in COLO-320 and SW-620 cells at 48hrs.

Comparable levels of inhibition in respect with a siRNA of control (Dharmacon) were achieved in

both types of cells. In spite of this, the biological effects were seen only in COLO-320 cells,

where the T-UCR is overexpressed about 2.5 times when compared with expression in normal

colon.

Figure S7. Downregulation by small interfering of uc.73A(P) induces apoptosis in COLO-

320 cells but not in SW-620 cells. Data obtained with caspase-3 assay in COL0-320 (upper

panel), where a significant increase in apoptotic cells is found, and in the control cells SW620

(lower panel), where no difference could be found. COLO-320 express high levels of uc.73A(P),

while in SW620 the expression is comparable with the normal colon levels.

III. Supplemental Tables

Table S1. T-UCRs expression in 22 human normal tissues (3 in duplicate, from 2 different

individuals).

Table S2. T-UCRs differentially expressed in CLL, CRC and HCC identified by ANOVA analysis

at P<0.005 (GeneSpring GX software).

Table S3. Genomic location of UCRs is correlated with CAGR.

(databases as in (Bejerano et al., 2004); (Calin et al., 2004))

Table S4. Negative correlations between the expression of miRNAs and T-UCRs in CLL

patients. All validated negative correlation by FDR method at 0.01 threshold, or 1% of false

positive results, and with an R correlation lower as 0.40 were considered.

IV. Supplemental References

Bejerano, G., Pheasant, M., Makunin, I., Stephen, S., Kent, W. J., Mattick, J. S., and Haussler,

D. (2004). Ultraconserved elements in the human genome. Science 304, 1321-1325.

Benjamini Y and Hochberg Y. Controlling the False Discovery Rate: a Practical and Powerful

Approach to Multiple Testing. Journal of the Royal Statistical Society. B, 57, 289–300.

Calin, G. A., Sevignani, C., Dumitru, C. D., Hyslop, T., Noch, E., Yendamuri, S., Shimizu, M.,

Rattan, S., Bullrich, F., Negrini, M., and Croce, C. M. (2004). Human microRNA genes are

frequently located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci

U S A 101, 2999-3004.