Background
In industrialized countries, about 30-50% of Hodgkin lymphomas (HLs) have been associated with the Epstein-Barr virus (EBV), but the impact of EBV infection on the clinical outcomes has been difficult to measure, because most HLs respond well to chemotherapy. In a multicenter retrospective survey, the prognosis was found to be worse for adult EBV
+ HLs than for their EBV
- counterparts [
1]. However, the underlying mechanism is still unknown.
In addition to HL, EBV is also associated with Burkitt's lymphoma, nasopharyngeal carcinoma, and other malignancies [
2]. Although EBV can switch its life cycle between a lytic phase and a latent phase [
3], the virus exists only in a latent phase in EBV-infected tumor cells. The latent phase is characterized by the variable expression of a limited set of virus-encoded genes, including 6 nuclear antigens (EBNAs 1, 2, 3A, 3B, 3C, & LP), 3 latent membrane proteins (LMPs 1, 2A, 2B), and 2 small homologous RNAs (
EBERs 1& 2). Depending on the expression patterns, the latent phase can be further classified into three types [
4]. EBNA1 and
EBERs are the only EBV-encoded genes common to all latencies. They are probably indispensable for latency maintenance or malignant transformation.
In the latency phase, EBNA1 maintains replication of the episomal form of the virus [
5], and it enhances the growth of HL cells [
6]. In contrast, the roles of
EBERs are unclear and controversial.
EBER probably interacts with both a ribosomal protein L22 and an RNA-dependent protein, PKR [
7]. According to one model, PKR may induce apoptosis;
EBERs antagonize PKR-mediated apoptosis, whereas L22 competes with PKR for
EBERs binding and abolishes the anti-apoptotic activity of
EBERs. The anti-apoptotic activity of
EBERs is consistent with the finding that EBV infection could reduce apoptosis in Burkitt's lymphoma [
8,
9]. In addition, PKR-independent anti-apoptotic activities of
EBERs have been reported [
10], but the mechanism and clinical significance are still unknown.
To address the mechanism and clinical significance of the anti-apoptotic activity of
EBERs, we analyzed the
EBER1-induced changes in HL cell lines using microarrays and found that
EBER1 suppressed
p21cip1/waf1 transcription.
p21cip1/waf1 is also known as the cyclin-dependent kinase inhibitor 1A (CDKN1A), and it normally causes cell cycle arrest at the G1/S phase, and induces or inhibits apoptosis [
11‐
13]. We demonstrated that decreased
p21cip1/waf1 transcription is associated with increased resistance to drug-induced apoptosis in HL cell lines. Most significantly from a clinical perspective, suppression of
p21cip1/waf1 and the increased resistance to drug-induced apoptosis are associated with a worse prognosis in cases of EBV
+ HLs.
Methods
Cell lines
KMH2 and L428, two EBV-negative HL cell lines, were obtained from the German Collection of Microorganisms and Cell Culture (DSMZ, Braunschweig, Germany). Similar to the classical Reed-Sternberg cells in HLs, these cell lines are CD30
+/CD15
+/CD3
-/CD19
-, and they have rearrangement of the immunoglobulin heavy-chain genes [
14]. These cell lines were cultured in RPMI1640 containing 10% fetal bovine serum, 50 μg/mL streptomycin, and 50 U/mL penicillin, at 37°C with 5% CO
2.
Construction of plasmids expressing EBER1 or antisense-EBER1 and selection of stable clones
The plasmid p9362 with an H1 promoter for transcription of small RNAs was used as the expression vector. The plasmid also expressed EGFP, as well as Kanamycin in bacteria or G-418 in eukaryocytes. The 167-nucleotide EBER1 or antisense-EBER1 was inserted into p9362 for construction of p9362-EBER1 or p9362-antisense-EBER1. Four cell lines were constructed: KMH2 transfected with p9362-EBER1 (KE), L428 transfected with p9362-EBER1 (LE), KMH2 transfected with p9362 (K9), L428 transfected with p9362 (L9). One additional control cell line, KMH2 transfected with p9362-antisense-EBER1 (K-anti-E), was also established.
Briefly, about 1 × 106 KMH2 or L428 cells were transfected with 30 μg EBER1-expressing plasmid (p9362-EBER1), antisense-EBER1-expressing plasmid (p9362-antisense-EBER1), or control plasmid (p9362) by electroporation with an ECM630 system (BTX, Holliston, MA). Stable clones were selected in RPMI1640 and 10% fetal bovine serum containing 1 mg/mL GENETICIN (Invitrogen, Carlsbad, CA). EBER1+ cell lines (KE & LE), an antisense-EBER1+ cell line (K-Anti-E), and plasmid-only cell lines (K9 & L9) were established. A purity of greater than 99% of EGFP+ cells was confirmed by flow cytometric analysis and expression of EBER1 was confirmed by Northern blotting.
Northern blotting
Northern blotting was done with dig-labeled probes: 5'-ACAGACACCGTCCTCACCACCCGGGACTTGTACCCGGGACGGGTG-3' for EBER1 or 5'-TCTTCTCTGTATCGTTCCAATTTTAGTATATGTGCTGCCG-3' for U6.
Briefly, 2.5 μg small RNAs were separated on a 5% denaturing polyacryamide gel and transferred to a Hybond-N membrane (Amersham, Little Chalfont, Bucks, UK). The membrane was hybridized with the EBER1 or U6 probe at a concentration of 50 ng/mL in a buffer containing 50% formamide at 52°C for 16 hours. The membrane was then washed twice with 2 × SSC in 0.1% SDS at 25°C for 5 min, and twice with 0.2 × SSC in 0.1%SDS at 68°C for 10 min. Anti-digoxigenin-AP and CSPD (Roche, Mannheim, Germany) were used for development of chemiluminescence.
Microarray
The Affymetrix chip, Human Genome U133 plus 2.0, was used to obtain genome-wide transcriptional profiles of the four stable cell lines (K9, KE, L9, and LE). First-strand cDNAs were synthesized from 10 μg of total RNAs with a T7-promoter-oligo (dT) primer. After second-strand synthesis, biotin-labeled cRNAs were transcribed from the T7 promoter. The cRNAs were fragmented into sizes ranging from 35 to 200 nucleotides, labeled with streptavidin-PE, mixed with control RNAs (bioB, bioC, bioD, and cre), and hybridized with the glass slides according to the GeneChip Expression Analysis Technical Manual from Affymetrix. The arrays were scanned with GenePix 4000B (Molecular Devices, Sunnyvale, CA, USA), and the data were extracted with Affymetrix Microarray Suite (MAS) software and submitted to Gene Expression Ominbus at
http://www.ncbi.nlm.nih.gov/geo/ with the accession number GSE12427.
Local normalization of the extracted raw data was done online at
http://pevsnerlab.kennedykrieger.org/snomadinput.html[
15]. In this method, the significance of the difference between the
EBER1+ cell lines and the control cell lines was designated by the z-score. For example, the
p21cip1/waf1 transcripts had a 2-fold decrease from 2260 in K9 to 1164 in KE, a mean level of 1712, and a z-score, Z
K, of - 2.5. The z-score meant that the 2-fold decrease for
p21cip1/waf1 was located at -2.5 standard deviations, when normalized with respect to the changes of genes with a similar mean level of expression. Each gene thus had a Z
K for KE
vs. K9, and a second Z
L for LE
vs. L9. The changes for the gene were concordant if both z-scores were positive or both z-scores were negative.
RT-PCR for p21cip1/waf1 splicing variants
Eight splicing variants of
p21cip1/waf1 have been reported: variant 1, variant 2, Alt-a, Alt-a', Alt-b, Alt-c, B, and C [
16,
17]. A universal RT primer, 5'-RS-CATTAGCGCATCACAGTCGC-3' (5506-5487), was used for converting mRNAs of all splicing variants into cDNAs. This RT primer consists of a
p21cip1/waf1 binding sequence tagged with a random sequence (RS): 5'-
GTATACTGCAGGGTCTGATAC-3'. A fluorescent universal reverse PCR primer, 5'-FAM-ATAGGTATACTGCAGGGTCTGATAC-3', and a forward PCR primer specific for each variant were then used to amplify the cDNAs. The specific forward primers were: 5'-CTGCCGAAGTCAGTTCCTTG-3' (variant 1, 18~37), 5'-ACTCAGAGGAGGTGAGAGAG-3' (variant 2, 79~98), 5'-GGTGGCTATTTTGTCCTTGG-3' (Alt-a, a' and b, - 835~-814), 5'-GGAGGCAAAAGTCCTGTGTT-3' (Alt-c, - 2219~-2197), 5'-AAGGAGGAGAGAGACCCT CT-3' (B, 5266~5286), and 5'-CTAGAAAATCCAGTTGCTG-3' (C, 3954~3972). RT-PCR for
β 2M was used as an internal control. A reverse primer, 5'-RS-CAGAATTTGGAATTCATCCAA-3', and a forward primer, 5'-CTTTGTCACAGCCCAAGATAG-3', were used. The numbers in parentheses are the locations of the primers, with respect to the genomic position of the
p21cip1/waf1, with the transcription start site of variant 1 being +1.
The PCR products were separated by high-resolution capillary electrophoresis and quantified by fluorescence. The size of the PCR products in base pairs were: 216 for variant 1, 296 for variant 2, 179 for alt-a, 225 for alt-a', 191 for alt-b, 349 for alt-c, 208 for B, and 213 for C. The 3 variants (alt-a, a' and b) from the same PCR had different sizes due to alternative splicing.
Real-time RT-PCR for p21cip1/waf1
Taqman® Gene Expression Assays (Applied Biosystems, Foster City, CA) were used for real-time RT-PCR of p21cip1/waf1 (HS_0001121172_m1) and actin (HS_00357333_g1). Total RNAs were extracted from K9, KE, L9, and LE cells. The RNAs were reverse transcribed into cDNAs with random hexamers. After an initial 10-min denaturation step at 95°C, 45 cycles of PCR were performed with denaturation at 95°C for 15 sec and annealing at 60°C for 1 min on the Stepone™ Real-Time PCR system (Applied Biosystems, Foster City, CA). The threshold cycle of p21cip1/waf1 minus that of actin was calculated.
Western blotting for p21cip1/waf1, EGR1, STAT1, p53, SirT5, GAPDH, cyclin-dependent kinases (CDKs) and cyclins
Western blotting was performed with the following antibodies: p21cip1/waf1 (clone CP74, Lab Vision, Fremont, CA), SirT5 (rabbit polyclonal, Abcam, Cambridge, UK), STAT1 (clone 42H3, Cell signaling, Danvers, MA), EGR1, p53, CDK6, and GAPDH (clone 588, clone DO-1, clone C-21, and clone FL-335, SANTA CRUZ, Santa Cruz, CA), CDK1, CDK2, CDK4, and cyclin B1 (clone POH-1, clone Poly6332, clone Poly6333, and clone Poly6334, Biolegend, San Diego, CA), cyclin A (clone 25/CyclinA, BD Pharmingen, Franklin Lakes, NJ), cyclin D2, and cyclin E (clone DCS-3.1, and clone HE12, Abcam, Cambridge, UK).
Cell cycle analysis
Cells were cultured in RPMI1640 medium plus 10% fetal bovine serum for 24 hours. The cells were fixed with 75% ethanol at -20°C overnight and were stained in 50 μg/mL propidium iodide (Sigma, Saint Louis, MO, USA), 0.05% Triton X-100, 0.1 μg/μL RNase A, and 1× PBS at 37°C for 30 min in the dark. The stained cells were washed with 3 mL PBS and suspended in 500 μL PBS for flow-cytometric analysis.
Apoptosis induced by TSA & MG115 measured by flow cytometry for Annexin V and propidium iodide
TSA (Trichostatin A, a histone deacetylase inhibitor from Sigma, Saint Louis, MO, USA) and MG115 (a proteasome inhibitor from Calbiochem, San Diego, CA) were used to induce apoptosis through up-regulation of p21cip1/waf1. About 1 × 105 KE, K9, LE, or L9 cells were grown in 1 mL of medium containing by 0.5 μM TSA or by 0.4 μM MG115. After 2 days, the cells were harvested, washed twice with 1× PBS, and suspended in 100 μL 1× binding buffer containing 10 mM Hepes at pH 7.4, 140 mM NaCl, and 2.5 mM CaCl2. The cells in 100 μL binding buffer were mixed with 5 μL APC-conjugated Annexin V (BD Pharmingen, Franklin Lakes, NJ) and 5 μL of 50 μg/mL propidium iodide at 25°C for 15 min. The stained cells were diluted with 400 μL of 1× binding buffer and analyzed by flow cytometry within 1 hour.
Induction of p21cip/waf1 by TSA or MG115 measured by ELISA
K9, KE, L9, LE cells were treated with 0.5 μM TSA or 0.4 μM MG115 for 1 day. Untreated cells were used as controls. The amounts of p21cip/waf1 were measured with an ELISA kit (Total p21cip/waf1 Sandwich ELISA Kit, Cell Signaling, Danvers, MA). The ratio of p21cip/waf1 in the treated cells to that in the untreated cells was calculated.
Briefly, after cell lysis and protein extraction, 25 μg proteins were loaded onto p21cip/waf1 antibody-coated microwells at 37°C for 2 hours. A detection antibody for p21cip/waf1, an HRP-linked secondary antibody, and the TMB substrate were applied sequentially. The absorbance at 450 nm was measured, and the background absorbance was subtracted out. The ratios of the absorbances of treated cells to those of untreated cells were calculated.
Cell growth curve
For growth curve, 1 × 104 K9, KE, L9, or LE cells were grown in 0.5 mL medium containing 0 to 0.5 μM TSA or 0 to 0.4 μM MG115 for up to 2 days. The cells were stained with 0.4% trypan blue and counted by hemocytometer.
Tissue samples
Biopsy specimens of 94 HLs with sufficient tissues and clinical data for further investigations were retrieved from the lymphoma database at the Department of Pathology of the National Taiwan University Hospital. The study was approved by the ethics committee of the National Taiwan University Hospital.
Immunohistochemistry for p21cip1/waf1 and in situ hybridization for EBER1
Immunoperoxidase stain for p21cip1/waf1, active caspase 3, and Ki-67 were performed on sections of formalin-fixed, paraffin-embedded HL cell blocks and tissue samples with the antibodies to p21cip1/waf1 (clone EA10, Calbiochem, Darmstadt, Germany), active caspase 3 (C92-605, BD Pharmingen, Franklin Lakes, NJ), and Ki67 (MIB-1, DAKO, Glostrup, Denmark). For each case, 50 Reed-Sternberg cells were examined, and the percentages of positive cells were recorded.
In situ hybridization for EBER1 was done on formalin-fixed, paraffin-embedded tissue sections with a dig-labeled probe, 5'-ACAGACACCGTCCTCACCACCCGGGACTTGTACCCGGGACGGGTG-3'. The sections were detected with immunoalkaline phosphatase and developed with BCIP/NBT.
Statistical analysis
The clinical data were extracted from the medical records. Two-sample comparisons were done with the Fischer's test for categorical data and the Mann-Whitney test for continuous data. 2-year overall survival (OS) rate and disease-free survival (DFS) rate analyses were done with the Kaplan-Meier method.
Discussion
We found that
EBER1 suppresses
p21cip1/waf1 transcription and inhibits drug-induced apoptosis, but does not change the cyclins [
31] except for cyclin D2. The anti-apoptotic activity of
EBER1 could be critical in the rescue of HL cells from apoptosis. These cells have crippled immunoglobulin genes and should have undergone apoptosis in the germinal center [
32]. The increased resistance to drug-induced apoptosis offers a possible explanation for the worse clinical behavior of EBV
+ HLs [
1].
Although most HLs respond to chemotherapy, about 25% of HLs are refractory or relapse after an initial response [
1]. For these cases, it is important to identify prognostic factors, such as the sites and extent of relapse, and to adjust treatment accordingly. Apart from these obvious clinical predictors, biological predictors, such as suppression of p21
cip1/waf1, should be useful. In fact, the clinical usefulness of p21
cip1/waf1 as a prognostic factor has been reported repeatedly for other tumors [
33], and our data suggested possible use of p21
cip1/waf1 as a prognostic factor in EBV
+ HL.
In our study, we have used TSA, a histone deacetylase inhibitor, and MG115, a proteasome inhibitor, to test the effect of
EBER1 on drug-induced apoptosis in HL cell lines, because similar drugs are currently being evaluated for clinical usages. Bortezomib is a proteasome inhibitor found to cause cell cycle arrest and to induce apoptosis in HL cell lines. However, in a pilot study, the drug demonstrated minimal activity in relapsed and refractory HLs [
26]. Similar to the conclusion of this pilot study, our data imply that treatment with proteasome inhibitors in EBV
+ refractory/relapsed HLs is likely to be ineffective. Vorinostat is a histone deacetylase inhibitor, which was found to suppress p21
cip1/waf1, cause cell cycle arrest, and induce apoptosis in HL cell lines [
27]. Because we have compared only
EBER1+ and EBER1
- HL cell lines, whereas the Reed-Sternberg cells in EBV
+ HLs also express EBNA1 and LMPs, it would be interesting to see whether these drugs are effective in EBV
+ HLs in future clinical trials.
From the perspective of tumor biology, there are 3 subtypes of latency in EBV-infected tumor cells. EBV
+ HL is typical of type II latency, and only a limited set of virus-associated genes is expressed. These genes include
EBNA1, LMP1 &
LMP2, and
EBER s. EBNA1 keeps the viral genome in an episomal form, LMP1 transmits CD40-like signals to compensate for the lack of B-cell receptors [
34], and LMP2A is a B-cell receptor mimic that is essential for survival [
35]. Because
EBER1 may suppress
p21cip1/waf1 transcription, and p21
cip1/waf1 is necessary for lytic induction by EBV-encoded ZTA [
36,
37],
EBER1 may be critical for the maintenance of the latency phase.
To integrate the data on apoptosis, drug-resistance, and maintenance of the latency phase, we have presented a model of
EBER1-induced p21
cip1/waf1 suppression through EGR1, and STAT1 (Fig
5D). Other genes in Table
1 might be related to this model too. Sir-2-like 5 (SirT5) is a histone deacetylase that could suppress
p21cip1/waf1 transcription [
23,
24]. HEXIM2, a double-stranded RNA-binding protein [
38], and MATR3, which processes double-stranded RNAs [
39], could have interacted with
EBER1. Finally, TRIM22 is important for antiviral defense [
40].
In addition to HL, Burkitt's lymphoma and post-transplantational lymphoproliferative disorder are also EBV-associated B-cell lymphomas that share a similar pathogenetic mechanism, in which EBV infection is important in the immortalization and transformation of B cells. In experimental conditions, EBV infection of primary B cells leads to p21
cip/waf1suppression and overrides genotoxin-induced G1 arrest [
41]. These EBV-infected primary B cells are in type III latency and the suppression of p21
cip/waf1 is due to post-transcriptional regulation, whereas HL is in type II latency and the suppression of p21
cip/waf1 is regulated at the transcriptional level. Despite of the differences, the suppression of p21
cip/waf1 appears to be a common event critical for the development of these lymphomas.
At greater than 5*10
6 copies per cell,
EBERs are the most abundant RNAs in EBV-infected cells. Although
EBERs seem to prevent apoptosis through interacting with PKR, the nuclear localization of
EBERs and cytoplasmic localization of PKR make a direct interaction unlikely [
10]. Recently,
EBERs were found to increase transcription or mRNA stability of IL-10, IL-9, or IGF1 in lymphoma or carcinoma cell lines [
42,
43]. With recognition of transcriptional regulation by
EBERs and other noncoding RNAs as important biological processes [
44], suppression of
p21cip/waf1 transcription deserves further investigations, because of the direct link with apoptosis and the known example of artificial microRNAs in regulating
p21cip/waf1 transcription [
45].
Diepstra et al [
46] reported a series of 412 HL patients with a median age of 35 years. In patients older than 50 years, the five-year failure free survival was 60% in EBV
+ cases
vs. 85% in EBV
- cases (
p = 0.01). Our series of 94 cases had a median age of 31 years. In patients older than 45 years, the 5-year disease free survival was 37% in EBV
+ cases
vs. 74% in EBV
- cases (
p = 0.02). In patients younger than 45 years, the 5-year disease free survival was 50% in EBV
+ cases
vs. 67% in EBV
- cases (
p = 0.17). Our data and those reported by Diepstra et al were consistent in showing that EBV
+ HL had a worse prognosis in the older age group.
In conclusion, EBER1 suppresses p21cip1/waf1 transcription and confers resistance to drug-induced apoptosis in HL cell lines. Biologically, this anti-apoptotic activity might be important in the rescue of Reed-Sternberg cells and in the maintenance of the latent phase. Clinically, the suppression of p21cip1/waf1 in EBV+ HL predicts a worse prognosis, and the possibility of increased resistance to drug-induced apoptosis might have therapeutic implications.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
TYL performed the most of experiments and wrote the manuscript. SJW analyzed the clinical data. MSH and FYL constructed the plasmids and the cell lines. MHT performed the bioinformatics on the microarray data. CHT established EBV-infected KMH2 and lymphoblastoid cell lines. SMH participated in drafting the manuscript. CWL coordinated the whole project and revised the manuscript. All authors read and approved the final manuscript.