Tumor suppressor genes FHIT and WWOX are deleted in primary effusion lymphoma (PEL) cell lines

Abstract

Primary effusion lymphoma (PEL) is a diffuse-large B-cell lymphoma with poor prognosis. One hundred percent of PELs carry the genome of Kaposi sarcoma-associated herpesvirus and a majority are coinfected with Epstein-Barr virus (EBV). We profiled genomic aberrations in PEL cells using the Affymetrix 6.0 SNP array. This identified for the first time individual genes that are altered in PEL cells. Eleven of 13 samples (85%) were deleted for the fragile site tumor suppressors WWOX and FHIT. Alterations were also observed in the DERL1, ETV1, RASA4, TPK1, TRIM56, and VPS41 genes, which are yet to be characterized for their roles in cancer. Coinfection with EBV was associated with significantly fewer gross genomic aberrations, and PEL could be segregated into EBV-positive and EBV-negative clusters on the basis of host chromosome alterations. This suggests a model in which both host genetic aberrations and the 2 viruses contribute to the PEL phenotype.

Figure 1

PCA of PEL. PCA shows the presence of 2 distinct clusters formed by endothelial cells in green and all the 17 (excluding JSC1) PEL cells in blue, irrespective of whether they were from culture or xenograft. The x-, y-, and z-axes represent PC1, PC2, and PC3, respectively, accounting for 14.9%, 12.2%, and 11.7% of variability in the data.

Figure 2

CFS tumor suppressor genes FHIT, WWOX, and GRID dot plot in PEL. Dot plot representation of markers distributed along the chromosome. (A,C) The loss of FHIT, WWOX, and GRID2 from chromosomes 3, 16, and 4, both in the 6.0 and 500K SNP array are shown. The markers (from each of the 13 PELs) are represented by black dots on a log₂ scale (amplification denoted by dots above and deletion by dots below the normal 0 line) with the cytoband at the base of plot. The dots identified by ∧ represent alterations that only occurred in 1 of 13 PELs and were thus considered exceptions. (B,D) Data represent 6 nonlymphoma tumor samples and 5 non-PEL lymphoma controls, respectively.

Figure 3

qPCR verification of CFS tumor suppressor gene loss in PEL. (A-E) The qPCR results for DERL1, FHIT, GRID2, WWOX, and KSHV, respectively, are shown. Shown is the stacked relative level (ddCT) for each gene on the vertical axis and the 2 classes of 13 non-PEL (other) and 13 PEL cell lines on the horizontal axis. The contribution of individual cell lines is indicated by the gray level. Because amplifications and deletions result in only a 2-fold change in signal in the case of cellular genes (and ∼ 50-fold for KSHV because there are ∼ 50 copies of the KSHV genome in each PEL cell) the stacked representation integrates both the degree of change as well as the number of cell lines that contribute to the signal in each group (a similar metric was previously validated).

Figure 4

Quantification of CNV in PEL, separated by EBV status. (A) There are significantly more amplifications in EBV-negative PELs than in EBV-positive PELs. (B) Although there is increased deletion of markers in the EBV-negative population, there difference is less conclusive. (C) PCA shows that PEL cells in culture form 2 distinct groups correlative with their EBV infection status: blue indicates EBV(−) and brown EBV(+) samples. The x-, y-, and z-axes represent PC1, PC2, and PC3, respectively, accounting for 18.9%, 18.1%, and 11.3% of variability in the data.