Early gene expression changes by Epstein-Barr virus infection of B-cells indicate CDKs and survivin as therapeutic targets for post-transplant lymphoproliferative diseases

Lymphoproliferative disease (LPD) associated with the B-cell tropic Epstein-Barr virus (EBV) causes significant morbidity and mortality in solid organ transplant (SOT)1 or bone mar- row transplant (BMT)2 recipients. The risk to develop EBV-LPD is greatest in EBV-seronegative transplant recipients, i.e., upon primary EBV infection, and in patients with exten- sive iatrogenic immunosuppression.3 EBV-associated B-cell lymphoma in the setting of post-transplant LPD (PTLD) is the leading malignancy in children after solid organ trans- plantation.4 Primary EBV infection of the EBV-seronegative allograft recipient may be acquired via the palatine tonsils, the natural portal of entry for the virus,5 or via EBV-infected donor-derived B-cells within the transplant.6 Thus, inexistent or heavily impaired pre-existing EBV-specific immune con- trol is an essential prerequisite for the pathogenesis of EBV- LPD that may emerge from recently EBV-infected B-cells derived from the natural portal of entry or from remote lym- phatic organs.

In vitro, EBV unfolds a unique capacity to growth-transform B-cells, both in mononuclear cells derived from EBV-sero- negative donors and in mononuclear cells from EBV-seropos- itive donors treated with immunosuppressants.7 Thus, inexistent or impaired cellular EBV-specific immune control, as in EBV-LPD, results in unrestricted proliferation of B-cells newly infected with EBV that, by default, establishes a latent infection.8 This suggests that similar pathogenic mechanisms may contribute to EBV-LPD in vivo and EBV-induced B-cell transformation in vitro.

Indeed, B-cells transformed by EBV in vitro have been instrumental in the study and characterization of the immune control of EBV, allowing demonstrating that control of latently EBV-infected B-cells by T-cells is essential. In this sense, B-cells transformed in vitro by EBV have been crucial for the engineering of EBV-specific cytotoxic T-cells (CTLs) to be used for adoptive transfer to patients to treat their EBV-LPD.9 Notably, adoptive transfer of CTLs has proven successful in fighting established EBV-LPD. Nevertheless, the complexity of manufacturing EBV-specific CTLs restricts their availability to few specialized centers. Other, more widely used treatments of post-transplant EBV-LPD include lowering of drug-induced immunosuppression or transfusion of antibodies against B-cells.10 Lowering immunosuppression, however, may not be applicable in BMT and is burdened with the risk of allograft rejection after SOT. Unspecific depletion of B-cells by anti-B-cell antibodies, on the other hand, may cause potentially deleterious B-cell deficiency. Also, prevention of EBV-LPD by, e.g., avoiding certain com- pounds for conditioning BMT recipients or antibodies against T-cells may be rather problem-prone, although it may con- tribute to improve the patients’ outcome.1 Thus, although prevention and treatment of post-transplant EBV-LPD focus on either specific or unspecific control of EBV, its manage- ment remains challenging.

A novel approach to fight post-transplant EBV-LPD could be targeting molecular events leading to transformation that are imposed on B-cells by the virus early after infection. B- cells from post-transplant EBV-LPD lesions express the same pattern of EBV genes (including EBNAs, LMPs and EBERs) observed during in vitro B-cell transformation by EBV lead- ing to the formation of lymphoblastoid cell lines (LCL).7 Notably, the efficiency of EBV to growth-transform B-cells in vitro using the classic growth-transformation assay may differ considerably depending on the B-cell status of differentia- tion.11 Thus, processes during EBV-induced B-cell transfor- mation in vitro may mirror the situation in the immunosuppressed transplant patient and their analysis including B-cells from distinct tissues could lead to the iden- tification of critical pathogenic steps that could be prophylac- tically or therapeutically targeted.

The aim of our work was to elucidate the early steps involved in EBV-induced B-cell transformation after infection with the virus. We used B-cells isolated from palatine tonsils from EBV-seronegative individuals. This enabled us to avoid using immune suppressants to counteract pre-existing EBV- specific CTLs and thus potential effects of immune suppres- sants on host gene expression. Furthermore, these samples enabled us to model the setting with the highest risk for the development of EBV-LPD, i.e., primary EBV infection. We used microarray technology to profile host gene expression to identify genes involved in EBV-mediated B-cell transforma- tion. Immunohistochemical analysis of post-transplant EBV- LPD confirmed the expression in patient samples and subse- quent functional experiments allowed the identification of potential host targets to efficiently counteract EBV-LPD.

Methods

Ethics statement

Our study was conducted according to the principles expressed in the Declaration of Helsinki. Written informed consent for the collection of samples and subsequent analysis was obtained from next of kin, caretakers or guardians on behalf of the minors/children participants involved in our study. The Institutional Review Board of the University Children’s Hospital of Zurich approved the study (IRB study StV29/06).

Isolations of B-cells

Human mononuclear cells were isolated from tonsils of donors undergoing routine tonsillectomy as previously described.12 B-cells were isolated using the B-cell isolation kit II according to the manufacturer’s instructions (Miltenyi Bio- tech, Bergisch Gladbach, Germany). Purity of the B-cells was above 95% in all the preparations.

EBV infection

EBV infection was performed as described earlier12 with minor modifications. Briefly, supernatants of B95.8 cells were centrifuged at 40,000g for 2 hr at 4◦C. The pellet was resus- pended in phosphate buffered saline and 10% of high-titer EBV was added to the cells.

EBER-1 in situ hybridization

The presence of EBER-1 in single cells was checked for by in situ hybridization and flow cytometry as reported,13 but with the modification that the EBER-1 hybridization probe was linked to Atto620 (Microsynth, Balgach, Switzerland).

Microarrays

Two samples, from two independent donors, were used per each experiment. Infected and noninfected purified B-cells were col- lected after 72 hr. RNA was isolated from the cells at the indicated time point by phenol-chloroform extraction using Trizol fol- lowed by additional column purification using the RNeasy mini kit (Qiagene, Hombrechtikon, Switzerland). Integrity of RNA was determined by Bioanalyzer (Agilent, Basel, Switzerland) and transcribed to cRNA using the Ovation Pico WTA system (NuGEN, Leek, The Netherlands). Hybridization on Affymetrix HG-U133 Plus 2 chip was performed at the Functional Genomic Centre Zurich. Data quality control and extraction were per- formed in the B-Fabric platform, which uses several R scripts based on bioConductor.14,15 RMA algorithm was used for sum- marization16 and standard RMA.17 Processed data were imported into the GeneGO portal for pathways enrichment analysis (Meta- Core from GeneGo, Encinidas, CA). Microarray data were sub- mitted to ArrayExpress (accession: E-MEXP-3582).

Cells treatments and analysis

Four LCLs were established from tonsillar mononuclear cells derived from four different EBV-negative donors, by infection with B95.8 EBV. Cells were kept in culture for 4–12 weeks. LCLs were split to a concentration of 0.3 3 106 cells/ml 16 hr before treatment; primary B-cells isolated from two donors were kept in culture 16 hr at 1.0 3 106 cells/ml in 100 ll volume. Increas- ing concentrations of flavopiridol, PD-0332991, YM155 (all from Selleck Chemicals, Houston, TX) and Terameprocol (Sigma-Aldrich, Buchs, Switzerland) dissolved 10 mM in di- methyl sulfoxide (DMSO) as a stock solution or corresponding maximum volume of DMSO as control were added to the cells.

Cells were collected after 24 and/or 48 hr and subjected to mea- surement of proliferation, cell cycle analysis, detection of apo- ptosis and Western blotting as follows: measurement of proliferation was performed using an MTS assay (CellTiter 96 AQueous One Solution Cell Proliferation Assay) according to the instructions of the manufacturer (Promega, Dubendorf, Swit- zerland). For cell cycle analysis, cells were harvested, washed in cold PBS and fixed in 70% ethanol. DNA was stained by incu- bating the cells in PBS containing propidium iodide (50 lg/ml, Sigma-Aldrich) and RNase A (100 lg/ml, Sigma-Aldrich) for 30 min at 37◦C. Detection of apoptosis was performed using a PE Annexin V Apoptosis Detection Kit I according to the instruc- tions of the manufacturer (Becton Dickinson, Allschwil, Swit- zerland). Data on cell cycle and apoptosis were collected using a fluorecence activated cell sorter FACSCanto II (Becton Dickin- son) and analyzed using FlowJo software. To perform Western blotting, cell extracts were prepared in Laemmli buffer (4% So- dium dodecyl sulfate, 20% glycerol and 120 mM Tris, pH 5 6.8). Electrophoresis and transfer to a membrane were performed using a NuPAGE system (Invitrogen). The antibodies used for immunoblots were rabbit anti-survivin (71G4B7, Cell Signaling Technology, Danvers, MA) and mouse anti-a-tubulin (DM1A, Sigma-Aldrich). For 5-(and-6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) labeling, isolated B-cells were stained with 5 lM CFSE (Sigma-Aldrich) for 5 min at room tempera- ture and then washed three times in PBS. CFSE-stained B-cells were infected with B95.8 EBV as already described and incubated at 37◦C in 5% CO2 air of relative humidity > 95% for indicated period. Then, the cells were subjected to flow cytometry using a FACSCanto II (Becton Dickinson) and the data were an- alyzed using FlowJo software.

Histology, preparation of tissue microarray and immunohistochemistry

Forty-eight biopsy specimens fixed in formalin and embedded in paraffin were collected and using a dedicated tool (micro- Arrayer, Beecher Instruments, Sun Prairie, WI) tissue microar- rays (TMAs) were constructed with 2 coring of 1-mm diameter from the blocks. From the obtained block new sections of 4 lm were cut (Supporting Information Fig. 6). Further details on patient samples and histological analysis can be found in Sup- porting Information Figs. 6 and 7 and Tables III and IV.

Statistical analysis

All statistical analysis was performed using Prism software (GraphPad Software) or Excel (Microsoft). Comparisons between groups were performed with an unpaired two-tailed Student’s t-test. p Values < 0.05 were regarded as statistically significant, unless otherwise stated.

Results

Cellular gene expression in B-cells after exposure to EBV

To elucidate the very early steps involved in EBV-induced B- cell transformation after cellular infection with EBV, we set out to profile cellular gene expression in B-cells derived from tonsils of EBV-seronegative individuals and newly infected with EBV ex vivo. Tonsils are considered as the portal of entry of EBV in the body. We chose 72 hr after infection as time point of analysis, because at this time infected B-cells start undergoing a phase of hyperproliferation lasting up to 3 days (Fig. 1a) that has been correlated with a phase of heightened oncogenic activity leading to B-cell transforma- tion.18 Knowing the proportion of EBV-infected B-cells at the time of analysis is important to judge the accuracy of gene expression analysis. By measuring EBER-1 expression, a surrogate marker for EBV infection, by flow cytometry we detected around 15% EBV positive B-cells after 72 hr (Fig. 1b and Supporting Information Fig. 1). Thus, 3 days postexpo- sure to EBV in vitro, up to one-fifth of the B-cells harbored EBV and would undergo hyperproliferation. Noninfected au- tologous tonsillar B-cells cultured for 72 hr were taken as baseline, and B-cell gene expression was assessed on an Affy- metrix Chip HG-U133 Plus 2 (EBV-infected samples n 5 2; mock-infected samples n 5 2).

Cyclin-dependent kinases essentially contribute to EBV- mediated B-cells transformation in vitro

We observed a clear and robust upregulation of genes associated with cell cycle or cell proliferation in B-cells after infection with EBV. Strikingly, cyclin-dependent kinases (CDKs) 1, 2, 4 and 6 were consistently upregulated in B-cells derived from tonsils 72 hr postinfection with EBV (Table 1 and Supporting Information Fig. 2). Upregulated expression of CDK1, CDK2 and CCNB1 was confirmed by real time quantitative polymerase chain reac-
tion (qPCR) (Supporting Information Fig. 5). Expression of CDKs is also higher in transformed LCLs, i.e., >30 days after B- cells infection with EBV, compared to primary B-cells as judged by microarray profiling, however, to a lesser extent than at 72 hr after infection (data not shown).

CDKs represent an appealing target for therapeutic inter- vention to eliminate EBV-mediated expansion of EBV- infected or EBV-transformed B-cells through a decrease of proliferation. Thus, to test the effect of specific CDK inhibi- tors on EBV-transformed B-cells, we treated four LCLs estab- lished from in vitro EBV-infected tonsillar B-cells, kept in culture for at least 30 days, with increasing concentrations of flavopiridol, a pan-CDK inhibitor, and PD0332991, a specific CDK4 and CDK6 inhibitor; both compounds are currently in phase II clinical trials.19,20 Cell viability assessed by MTS assay 48 hr after treatment indicated that all LCLs were sen- sitive to flavopiridol because viability dropped to 20% of con- trol (Fig. 2a). PD0332991, however, was not as effective in diminishing LCL viability reaching only 80% of control at the highest concentration of 1 lM (Fig. 2b). To verify whether the reduction in cell viability was owing to activation of apoptosis, we analyzed Annexin V exposure after 48 hr treatment with 1 lM flavopiridol. Indeed, more than 60% of the cells stained positive for Annexin V, indicating an activa- tion of apoptotic response (Fig. 2c). Survival of EBV-negative B-cells was not affected by flavopiridol (Fig. 2d) or by PD0332991 (Fig. 2e) even at the higher concentrations. Cell cycle distribution analysis of LCLs treated with increasing concentrations of flavopiridol showed a large increase of the sub-G1 population upon treatment with 1 lM flavopiridol at 48 hr (Fig. 2f), confirming that apoptosis might be the main mechanism responsible for flavopiridol-induced reduction of cell viability. Further, cell cycle analysis of LCLs treated with 1 lM PD0332991 showed an accumulation in G1 and decrease in S and G2 phases suggesting G1 arrest (Fig. 2f) with a rather modest impact on cell survival. Taken together, these results indicated that flavopiridol is more effective than PD0332991 in reducing viability of LCLs, and suggested that targeting CDKs, particularly CDK1, might be an effective approach to control and eliminate EBV-transformed B-cells.

EBV infection increases expression of antiapoptotic genes in tonsillar B-cells

EBV is known to inhibit apoptosis during latent infection.21 Thus, we aimed to identify additional apoptosis-related genes that are affected during the early events of EBV-mediated B- cell transformation. Indeed, BIRC5, coding for the strong antiapoptotic protein survivin,22 showed the highest upregu- lation (17-fold) among the ten most highly upregulated apo- ptosis-related genes (Table 2). BIRC5 (survivin) is a target of CDK1 and cyclin B1 (CCNB1) phosphorylation,23,24 both of which are also upregulated upon EBV infection (Table 1 and Supporting Information Fig. 3). We confirmed 4.5-fold up- regulation of BIRC5 expression by EBV infection in tonsillar B-cells using real-time qPCR (Supporting Information Fig. 5). Taken together, these results suggest that CDK1- CCNB1 and survivin represent an important signaling axis for the survival of EBV-infected B-cells.

Survivin is rather consistently expressed in EBV-positive post-transplant EBV-LPD

To investigate whether survivin is also expressed in EBV-pos- itive B-cell lymphomas in vivo, we generated TMA from a panel of 30 post-transplant EBV-LPD (Supporting Informa- tion Figs. 6 and 7 and Table 3). Presence of EBV was ascer- tained by EBER in situ hybridization (Fig. 3a) and by LMP1 staining (Supporting Information Fig. 7n). Expression of sur- vivin was assessed by IHC, and detection in more than 30% of the EBV-LPD cells was defined as positive (Fig. 3b). Sev- enty-one percent (n 5 17) of the evaluable EBV-positive tumors expressed survivin (Supporting Information Table 4). These results documented that survivin is consistently upreg- ulated in EBV-infected cells and indicated that survivin may have and important impact on B-cells transformation and proliferation in vivo when the immune system is compromised.

Survivin inhibition decreases survival of EBV-transformed B-cells

To verify whether survivin expression is important for sur- vival and proliferation of in vitro EBV-transformed B-cells, we tested the effect of two inhibitors of survivin, YM155 and terameprocol, on four freshly established LCLs. Both com- pounds act at the transcription level by blocking Sp1 activity on the survivin promoter.25 Cells were incubated with increasing concentrations of YM155 or terameprocol and cell proliferation was monitored after 48 hr by MTS assay (Fig. 4). Treatment with increasing concentrations of YM155 resulted in a significant decrease of LCL viability (Fig. 4a). One nanomolar YM155 reduced cell viability by 20% and 10 nM YM155 reduced cell viability to less than 20%. At the highest concentration tested (1 lM YM155) all cells were dead. The estimated IC50 for YM155 was 2.6 nM. Likewise, the treatment with increasing concentrations of terameprocol resulted in decreased cell viability, with an estimated IC50 of 46.4 lM (Fig. 4b), in good agreement with other reports.26 Analysis of preapoptotic cells exposing Annexin V confirmed induction of apoptosis in 30% of the cells by treatment with 10 nM YM155 and up to 70% with 100 nM YM155 or 100 lM terameprocol (Fig. 4c). It is noteworthy that treat- ment with 10 nM YM155 resulted in 80% decrease of cell vi- ability, but only 30% of the cells were positive for Annexin V, suggesting a differential impact of YM155 on proliferation and apoptosis. Survival of EBV-negative B-cells was not affected by treatment with YM155 (Fig. 4d) or terameprocol (Fig. 4e). Only at the highest YM155 concentration of 1 lM cell viability was reduced to 80%, but at the YM155 IC50 con- centration of 2.6 nM the viability of EBV-negative B-cells was not affected. Successful suppression of survivin expres- sion was verified at the mRNA level by qPCR (Supporting Information Fig. 8) and at the protein level by Western blot- ting (Fig. 4f). A significant decrease of mRNA expression was observed after 24-hr treatment with 10 or 100 nM YM155. Survivin protein decrease was evident by Western blotting af- ter 24-hr treatment with 100 lM terameprocol and after 48- hr treatment with 10 and 100 nM YM155 (Fig. 4f). These results indicated that YM155 is more efficient than terame- procol in decreasing survivin levels, and that treatment of EBV-transformed B-cells with survivin inhibitors results in induction of apoptosis and consistent decrease in cell survival.

Discussion

LPD associated with the B-cell tropic EBV causes significant morbidity and mortality in BMT2 or SOT recipients.1 An improved understanding of the early molecular events after EBV infection leading to B-cell transformation is essential to gain insight into LPD pathogenesis and help devise new ther- apeutic approaches. Here, we have analyzed the impact of EBV infection on early host gene expression in B-cells puri- fied from tonsils, the portal of entry for EBV in the body. We investigated B-cells from EBV-negative individuals to mimic molecular events occurring during primary EBV infec- tion of EBV-negative transplant recipients, the subpopulation with the highest risk for post-transplant EBV-LPD. Among the upregulated genes we identified in particular CDKs (CDK1/2/4/6), CCNB1 and the CDK1/CCNB1 target BIRC5 (survivin), an inhibitor of apoptosis, which we found highly expressed in established LCLs and in the majority of EBV- LPD tumors. Given that specific survivin inhibitors efficiently counteracted proliferation of EBV-transformed B-cells, CDK and survivin represent potential novel therapeutic targets for EBV-LPD.

Cellular transformation induced by EBV infection has been subject of active investigation by many groups. The mechanisms underlying cellular transformation by EBV have been investigated either by using wild-type virus,18,27–30 recombinant virus mutated in one or more genes31–33 or by ectopic expression of single EBV proteins known to be expressed in EBV-transformed cells (e.g., EBNA134,35 and EBNA236–39). Gene expression profiling of infected primary B-cells or of modified B-cell lines has revealed a number of genes whose expression is directly or indirectly influenced by
EBV proteins. It is not surprising that by using different profiling methods not always overlapping results have been obtained. In addition, differential activity in different cellular backgrounds has been shown, e.g., for EBNA1.40 However, some patterns seem to emerge as to the impact of EBV infec- tion on cellular gene expression, which are congruent with our results. EBNA2 is the first EBV gene to be detected after infection and at 72 hr postinfection we previously measured expression of most latency-associated EBV genes,12 in good agreement with that we found differential expression of vali- dated EBNA2 targets such as CDK2, 4 and 6,41 and EBNA1 target genes such as BIRC5.34 The fact that we find no EBNA3 or LMP1 target genes suggests that the impact of these proteins on cellular gene expression is less pronounced at 72 hr after infection. Indeed, LMP1 activation of NFjb signaling is surprisingly delayed during primary EBV infec- tion of B-cells.30

Our results, indicating an important role of cell cycle reg- ulators early during EBV infection and transformation, in particular of CDKs 1, 2, 4 and 6, urged us to test the impact of their inhibition on LCLs. Of the two inhibitors tested, fla- vopiridol, a pan-CDK inhibitor, was more effective than PD0332991, specific for CDK4 and CDK6, suggesting that CDK1 or CDK2 activity is more important for survival of EBV-transformed B-cells and might be more effective than CDK4 or CDK6 in EBV-LPDs. This is in good agreement with the decreased expression of CDK4 and CDK6 and the sustained overexpression of CDK1 and CDK2 LCLs observed 30 days after infection (M. Bernasconi, unpublished observa- tion). LCLs sensitivity to flavopiridol might also be explained by inhibition of CDK9-cyclin T1 (P-TEFb)42 and subsequent decreased activation of EBV promoter Cp and EBV latent gene EBNA1 and EBNA2 transcription, as shown in Burkitt’s lymphoma cells,43 might lead to decreased survival and offer a second layer of specificity in EBV-PTLD.

The importance of CDK1 in EBV-driven B-cell transfor- mation is also supported by the identification of survivin, a CDK1-cyclin B target, as an important antiapoptotic factor upregulated early upon EBV infection. The complex of CDK1-cyclin B phosphorylates survivin at Thr34. In the ab- sence of Thr34 phosphorylation survivin is rapidly degraded resulting in disruption of survivin-caspase 9 complexes lead- ing to increased caspase 9-dependent apoptosis in the M phase.23,24 Thus, in addition to CDK inhibitors, inhibitors of survivin expression could also represent important therapeu- tic compounds for the treatment of early and more advanced stages of EBV-LPD. Importantly, survivin expression is increased in LCLs when compared to EBV-negative B-cells. Indeed, inhibition of survivin expression using YM155 or ter- ameprocol was effective in killing LCLs, a model for post- transplant EBV-LPD, suggesting an important involvement of survivin in EBV-driven B-cell transformation. Interestingly, lower concentrations of YM155 (10 nM) significantly reduced LCLs proliferation without inducing apoptosis, whereas higher concentrations (>100 nM) strongly induced apoptosis suggesting a dual function, proliferative and antiapoptotic, of survivin in LCLs. Notably, survivin plays a significant role in other EBV-associated malignancies such as gastric carci- noma44 and NK/T cell lymphoma.26 Increased expression of survivin upon EBV infection in B-cells and in post-transplant EBV-LPD is novel. The collaboration of latent EBV protein EBNA1 with Sp1 to increase survivin expression by binding to the survivin promoter might provide a mechanistic expla- nation for upregulation of survivin in EBV-positive tumors.34 The importance of survivin in PTLPD is further highlighted by the fact that also EBV-negative DLBCL express survivin, as shown here and by others,45 indicating that additional mo- lecular mechanisms might be responsible for increased survi- vin expression in EBV-PTLD, usually arising within 1 year of transplant, and other PTLD arising later after organ trans- plantation. In addition, given the importance of EBV infec- tion in pediatric diffuse large B-cell lymphomas,46 it is imperative that therapies designed to target cell cycle and survivin are considered for EBV-positive PTLDs, which are often resistant to the current therapies that consist in a com- bination of chemotherapy (CHOP) and rituximab, a chimeric anti-CD20 monoclonal antibody.47

Collectively, in view of the suppressive effect of CDK and survivin inhibitors in vitro on proliferation and survival of LCLs documented here, it is tempting to propose studies in high-risk patients for LPD like EBV-seronegative transplant recipients. Notably, YM155 has shown a good pharmacologi- cal profile in clinical trials,48 and flavopiridol is currently in phase II clinical trial for relapsed mantle cell lymphoma and DLBCL (NCT00445341). Given that lowering of immunosup- pression is not possible in BMT recipients and administration of radio/chemotherapy with or without anti-B-cell antibodies puts the transplant at risk, our findings provide the rationale for additional studies on the role of EBV on survivin and CDKs regulation and malignant cells survival in PTLD that might lead to tailored interventions based on specific survivin and/or CDK inhibitors such as YM155 and flavopiridol.