Identification of genes with a correlation between copy number and expression in gastric cancer

Abstract

Background: To elucidate gene expression associated with copy number changes, we performed a genome-wide copy number and expression microarray analysis of 25 pairs of gastric tissues.

Methods: We applied laser capture microdissection (LCM) to obtain samples for microarray experiments and profiled DNA copy number and gene expression using 244K CGH Microarray and Human Exon 1.0 ST Microarray.

Results: Obviously, gain at 8q was detected at the highest frequency (70%) and 20q at the second (63%). We also identified molecular genetic divergences for different TNM-stages or histological subtypes of gastric cancers. Interestingly, the C20orf11 amplification and gain at 20q13.33 almost separated moderately differentiated (MD) gastric cancers from poorly differentiated (PD) type. A set of 163 genes showing the correlations between gene copy number and expression was selected and the identified genes were able to discriminate matched adjacent noncancerous samples from gastric cancer samples in an unsupervised two-way hierarchical clustering. Quantitative RT-PCR analysis for 4 genes (C20orf11, XPO5, PUF60, and PLOD3) of the 163 genes validated the microarray results. Notably, some candidate genes (MCM4 and YWHAZ) and its adjacent genes such as PRKDC, UBE2V2, ANKRD46, ZNF706, and GRHL2, were concordantly deregulated by genomic aberrations.

Conclusions: Taken together, our results reveal diverse chromosomal region alterations for different TNM-stages or histological subtypes of gastric cancers, which is helpful in researching clinicopathological classification, and highlight several interesting genes as potential biomarkers for gastric cancer.

Figure 1

DNA copy number change profiles in the 27 pairs of gastric samples. (A) Summary of chromosomal aberrations was shown. (B) The CNVs frequency of the whole genome was analyzed by aCGH. Gains were marked in red and losses in green.

Figure 2

Copy number aberration profiles of different gastric cancer subtypes revealed significantly altered genomic regions. The clinical sample group comparisons were performed for PD (n = 10) vs. MD (n = 11) (A), T1-2 (n = 11) vs. T3-4 (n = 13) (B), and N0 (n = 14) vs. N1-3 (n = 11) (C).

Figure 3

The mRNA level ofC20orf11in normal gastric tissues (N) and gastric cancer tissues (T). Relative expression was from expression microarray and qRT-PCR. Data were presented as Box and Whisker plots with *P < 0.05, **P < 0.01, and ***P < 0.001. ΔCt-values denoted up-regulated expression in T as compared with N. MD-amp, the MD gastric cancer samples with amplification at C20orf11; MD-normal, the MD gastric cancer samples without CNVs at C20orf11.

Figure 4

Correlation between copy number ratios and expression ratios in representative genes was validated by qRT-PCR. Each dot indicates an individual sample. The X axis displays copy number log ratios and the Y axis shows gene expression log ratios from microarray (A) or qRT-PCR (B). FDR, false discovery rate.

Figure 5

Correlation between copy number gains and overexpression of 31 genes at 8q11.1-q24.3. The genes were arranged in chromosomal order (p → q). An unsupervised hierarchical clustering of 50 gastric tissues with 32 genes overexpressed in gastric cancer samples (T) and underexpressed in matched adjacent noncancerous samples (N) revealed two distinct clusters (on the right). The genes included MCM4, PRKDC, UBE2V2, LYPLA1, LYN, CHCHD7, RAB2, GGH, SULF1, LACTB2, SLC10A5, PTDSS1, INTS8, LAPTM4B, PPM2C, RPL30, ATP6V1C1, SLC25A32, YWHAZ, ZNF706, CTHRC1, ENY2, RAD21, EIF3S3, MAL2, FAM91A1, FBXO32, SQLE, MYC, GPR172A, SIAHBP1, and GRINA (lack of aCGH data).

Figure 6

Concordant deregulation of candidate genes at 8q11.21 and 8q22.3. Expression of candidate genes located adjacent to MCM4 at 8q11.21 (A) and YWHAZ at 8q22.3 (B) shows the same trends as that of MCM4 and YWHAZ, respectively. (a) Schematic genome structures of neighboring candidate genes at 8q11.21 and 8q22.3; (b) Pearson’s correlation coefficients of expression log ratios among candidate genes; (c) Expression log ratio of the individual gene in each sample. Gain (−)/Gain (+) was referred to samples with or without gain at 8q11.21 or 8q22.3; Amp (+) was referred to samples with amplification at 8q11.21 or 8q22.3.