PRDM1 is a tumor suppressor gene in natural killer cell malignancies

Abstract

Natural killer cell lymphoma (NKCL) constitutes a rare and aggressive form of non-Hodgkin lymphoma, and there is little insight into its pathogenesis. Here we show that PRDM1 is a tumor suppressor gene in NKCLs that is inactivated by a combination of monoallelic deletion and promoter CpG island hypermethylation. We observed monoallelic deletion of PRDM1 loci in 8 of 18 (44%) NKCL cases. The other allele showed significant promoter methylation in 12 of 17 (71%) cases. In support of its role as a tumor suppressor gene, the reconstitution of PRDM1 in PRDM1-null NK cell lines led to G2/M cell cycle arrest, increased apoptosis, and a strong negative selection pressure with progressive elimination of PRDM1-expressing cells, which was enhanced when IL-2 concentration is limiting. We observed a progressive increase in PRDM1 expression--in particular, PRDM1α--in normal NK cells in response to IL-2 and in normal NK cells activated with an engineered NK cell target, K562-Cl9-mb21, suggesting its role in NK cell homeostasis. In support of this role, knockdown of PRDM1 by shRNA in normal NK cells resulted in the positive selection of these cells. We identified MYC and 4-1BBL as targets of PRDM1 in NK cells. Disruption of homeostatic control by PRDM1 may be an important pathogenetic mechanism for NKCL.

Fig. 1.

PRDM1 is silenced through a combination of deletion and promoter methylation. (A) NKCL cases with or without deletion is determined by the PRDM1/RPL13A ratio (three replicates), and cases with <0.75 ratio (dashed horizontal line) to that of human tonsil DNA are defined as deleted. The deletion status of the NKCL cases by aCGH is indicated by + and −. (B) PRDM1α promoter CpG island is hypermethylated in NK cell lines and NKCL cases. Methylation percentage (three technical replicates) of CpG dinucleotides upstream of the PRDM1α TSS is indicated as a heat map. Resting and activated NK cells (IL-2 for 7 d) are used as negative control samples. (C) qRT-PCR was performed on seven NKCL cases using primers specific for PRDM1 (α + β) (Upper) and PRDM1α (Lower). RPL13A was used for normalization. The deletion or methylation status of PRDM1 for each patient is shown. The expression level is expressed as a ratio of the level for resting NK cells. Data are mean ±SD. Resting human NK cells and activated NK cells derived from 14 d coculture with K562-Cl9-mb21.

Fig. 2.

Reconstitution of PRDM1 in PRDM1-null NK cell lines induces apoptosis and cell cycle arrest. (A) PRDM1 induces apoptosis in NK cell lines. Transduced, unsorted NK cells were stained with Annexin V-PE and tested with FACS. The rate of apoptosis was determined by measuring the proportion of Annexin V(+) cells in the GFP(+) population 2 d posttransduction. (B) Quantification of the rate of apoptosis in GFP(+) population of vector- or PRDM1-transduced cells. (C) Cell cycle profile of vector- or PRDM1 transduced KHYG1 and KAI3 cells. (D) The change in each cell cycle phase is calculated as PRDM1 − PMIG 3 d posttransduction.

Fig. 3.

Reconstitution of PRDM1α exerts negative selection pressure in malignant NK cell lines, which increases with decreasing doses of IL-2 in the culture medium. The FACS profile showing the percentage of GFP(+) cells of the vector- or PRDM1-transduced KHYG1 (A) and NKYS cells (B) before and after culturing with progressively decreasing doses of IL-2. (C) Comparison of the percentage of GFP(+) cells before and after treatment of PRDM1-transduced KHYG1 and NKYS cells with limiting doses of IL-2. Each data point was calculated as follows: [%GFP(PRDM1)]/[%GFP(vector)]. IL-2 posttreatment values were normalized to the values 48 h posttransduction. Data are mean ± SD of two independent experiments. (D) Comparison of apoptosis with limiting doses of IL-2 in the presence and absence of PRDM1. PRDM1- or vector-only transduced KHYG1 cells were treated with progressively decreasing doses of IL-2 for 2 or 5 d 48 h posttransduction. (E) PRDM1-dependent induction of apoptosis is enhanced with limiting IL-2 concentrations in malignant NK cells.

Fig. 4.

Knockdown of PRDM1 with shRNA results in the positive selection of human primary NK cells. (A) PRDM1 shRNA was PCR cloned inside the miR-30a backbone in MSCV-TMP. The percentage of GFP(+) cells was compared between vector and PRDM1 siRNA-transduced NKYS (B) or primary NK cells (C) 3 and 7 d (NKYS) or 4, 7, and 10 d (primary NK cells) posttransduction. Data are mean ± SD of two independent experiments.

Fig. 5.

PRDM1 regulates expression of genes involved in cell cycle and activation in NK cells. (A) qRT-PCR results of target genes regulated by PRDM1 in vector- or PRDM1-transduced, GFP-sorted KHYG1 cells 2 d posttransduction. (B) qRT-PCR results on vector- or PRDM1 shRNA-transduced, GFP-sorted NKYS cells 6 d posttransduction. (C) MYC mRNA expression was performed by qRT-PCR on primary NK cells activated by IL-2. RPL13A was used for normalization (n = 2).