Cytotoxic drug sensitivity of Epstein-Barr virus transformed lymphoblastoid B-cells

Background Epstein-Barr virus (EBV) is the causative agent of immunosuppression associated lymphoproliferations such as post-transplant lymphoproliferative disorder (PTLD), AIDS related immunoblastic lymphomas (ARL) and immunoblastic lymphomas in X-linked lymphoproliferative syndrome (XLP). The reported overall mortality for PTLD often exceeds 50%. Reducing the immunosuppression in recipients of solid organ transplants (SOT) or using highly active antiretroviral therapy in AIDS patients leads to complete remission in 23–50% of the PTLD/ARL cases but will not suffice for recipients of bone marrow grafts. An additional therapeutic alternative is the treatment with anti-CD20 antibodies (Rituximab) or EBV-specific cytotoxic T-cells. Chemotherapy is used for the non-responding cases only as the second or third line of treatment. The most frequently used chemotherapy regimens originate from the non-Hodgkin lymphoma protocols and there are no cytotoxic drugs that have been specifically selected against EBV induced lymphoproliferative disorders. Methods As lymphoblastoid cell lines (LCLs) are well established in vitro models for PTLD, we have assessed 17 LCLs for cytotoxic drug sensitivity. After three days of incubation, live and dead cells were differentially stained using fluorescent dyes. The precise numbers of live and dead cells were determined using a custom designed automated laser confocal fluorescent microscope. Results Independently of their origin, LCLs showed very similar drug sensitivity patterns against 29 frequently used cytostatic drugs. LCLs were highly sensitive for vincristine, methotrexate, epirubicin and paclitaxel. Conclusion Our data shows that the inclusion of epirubicin and paclitaxel into chemotherapy protocols against PTLD may be justified.


Background
Development of malignant B-cell lymphomas after organ transplantation is a significant complication arising as a side effect of the immunosuppression required for successful graft survival. The oncogenic Epstein-Barr virus (EBV) is the etiologic agent in the posttransplant lymphoproliferative disorder (PTLD) and AIDS related immunoblastic lymphomas (ARL) [1]. The reported overall mortality for PTLD often exceeds 50% [2,3]. The prognosis for PTLDs occurring after bone marrow transplantation is even worse [4,5]. Male patients with the rare inherited X-linked lymphoproliferative syndrome, showing specific immune defect against EBV infection, also often succumb to EBV induced malignant lymphomas [6].
EBV is a ubiquitous human herpesvirus that persists for life. Primary EBV infection can lead to mononucleosis (IM) in adolescence and in adults, manifested by a massive expansion of B cells. EBV-encoded transformationassociated proteins drive the proliferation of B lymphoblasts in IM, in PTLDs and in immunodeficiency syndrome-associated immunoblastic lymphomas. The EBV transformed cells express nine latency-associated viral proteins: EBNA1-6, LMP-1, -2A and -2B. This latency program is referred to as the type III latency. The same latency program is present in the in vitro proliferating lymphoblastoid cell lines (LCLs), generated by infection of normal human B cells with EBV. The fraction of B cells that is susceptible to in vitro transformation can be anything between 10% and 100% [7].
EBV drives the proliferation of human B cells in vitro and during primary infection in vivo. Strong T cell-mediated immune responses have been documented against EBV encoded latent proteins and a wide range of HLA class I molecules with EBV originated peptide epitopes have been identified [8][9][10]. EBV associated lymphoproliferative disease can develop only in the absence of a competent cytotoxic T cell immune surveillance. EBV associated lymphoproliferative disease may disappear upon treatment restoring the immune response against EBV-infected B cells.
Historically LCLs were often regarded as non-tumorigenic in immunosupressed mice upon subcutaneous inoculation, especially in comparison with highly tumorigenic Burkitt's lymphomas. However intraperitoneal inoculation regularly leads to development of generalized lymphomas with multiorgan involvement. SCID mice inoculated intraperitoneally with peripheral blood lymphocytes (PBL) from EBV-seropositive donors or with human LCLs, develop EBV-induced human lymphoproliferations within a few weeks. These lymphomas are classified as immunoblastic lymphomas, often with plasmacytoid features [11]. Histologically the PBL derived human-SCID tumors very much resemble the EBV positive large cell lymphomas of immunosuppressed patients [12]. The tumors of the immunocompromised patients or the experimental tumors growing in immunodefective mice as well as the in vitro growing LCLs show very similar phenotypes. All three express the same spectrum of cell surface markers, B cell activation antigens and adhesion molecules. All three have normal karyotype and show identical viral gene expression patterns.
The risk of PTLD has been found to depend upon the type of the transplanted organ, the immunosuppressive regimen, the age, the underlying illness and the EBV status of the recipient at the time of transplantation. The estimated incidence of PTLD ranges from 1-4% after renal transplantation to 19% after intestinal transplantation. In bone marrow allograft recipients PTLD is relatively uncommon (1%) [13,14] except for when certain high risk regimes, such as in vitro T-cell depletion (TCD) are used, when the risk may rise to 30% [4,15] PTLD following allogenic stem cell transplantation usually derives from donor lymphocytes. The risk of PTLD is greater if the host is EBV-seronegative at the time of transplantation and/or if there is a mismatch between the donor and recipient HLA types [1].
No controlled studies have been performed in the management of PTLD and most of the recommendations for therapy come from small cohorts at single institutions [1]. The relative importance of T cell impairment, EBV and clonal proliferation has led to the following strategies: reduction of immunosuppression or prophylactic restoration of T-cell immunity [16], antiviral therapy and chemotherapy. Reducing the immunosuppression leads to complete and durable remission of PTLD [17] for 23-50% of patients after organ transplantation (SOD) but will not be efficient in the BMT setting. Reduction of immunosuppression is frequently the first therapeutic step, and patients who have had organ rejection have a much poorer prognosis.
For secondary EBV lymphomas post BMT none of the above regimens will suffice, except when specifically restoring EBV specific immunity [16]. Anti-B-cell monoclonal antibodies are an effective therapy for PTLD. A combination of anti CD21 and anti-CD24 antibodies were used. However these antibodies are not commercially available and interest has therefore recently turned towards Rituximab, a humanized monoclonal antibody against the pan-B cell marker CD20. The response rate of Rituximab treatment is 65% with a relapse rate of 18% [18,19].
In cases that are not responding to the above mentioned therapy, chemotherapy is used as the second or third line of option.
PTLD patients are exposed to a high mortality risk due to chemotherapy treatment related toxicity [1].
The side effects of a drug are the limiting factor in determining the dose. Increasing the dose of a drug leads to a more frequent appearance of side effects. Chemotherapeutic agents can also compromise the survival of the graft. Drugs that are effective even at low doses could therefore, not only give fewer side effects, but also provide a better graft survival.
The aim of this study was to find the most effective cytotoxic drugs against EBV-induced lymphoproliferations. We carried out over 6500 cell survival and proliferation assays, on single cell level, in order to establish the pattern of in vitro efficacy of the most commonly used cytotoxic drugs against EBV-transformed lymphoblastoid cells. During the time of experiment the LCLs were cultured in IMDM (Sigma) supplemented with 20% FCS (Sigma), 100 mmol L-glutamine (Sigma) and 80 µg/ml gentamicin (Sigma). Cell suspensions were grown in a humidified incubator at 37°C in an atmosphere containing 5% CO 2 . Cell counts were adjusted to an optimal concentration of 1 × 10 6 cells/ml and the cells were fed twice a week. The absence of mycoplasma contamination was assured by regular monitoring with Hoechst 33258 staining.

Drugs
For the in vitro drug sensitivity test 29 drugs were used (summarized in Table 1). All the drugs were dissolved in 50% dimethyl sulfoxide (DMSO) and printed on 384 well plates, using a high density metal pin array (with 50 nl replica volumes) in Biomek 2000 fluid dispenser robot (Beckman). The same robot was used to generate the drug masterplates containing the triplicates of four different drug dilutions using a single tip automatic dispenser head.
The highest drug concentration was selected as the physicochemical maximally achievable drug concentration at 600 times dilution (50 nl drug in 30 µl suspension assay volume) The Ratio of Maximum Achieved Plasma Concentration (RMAPC) was determined for each drug concentration in order to compare the in vitro tested concentrations to the in vivo maximally achieved plasma concentration (Cmax) levels.
RMAPC was calculated according to the following equation:

In vitro drug sensitivity assay
In vitro drug sensitivity of LCLs was assessed using a 3-day cell culture on microtiter plates. 28 drugs were tested, each at 4 different concentrations in triplicates on 384-well plates (Greiner). Each well was loaded with 30 µl cell suspension containing 3000 cells. After three days of incubation the live and dead cells were differentially stained using fluorescent VitalDye (Biomarker, Hungary). The exact number of live and dead cells was determined using a custom designed automated laser confocal fluorescent microscope (a modified Perkin-Elmer UltraView LCI system) at the Karolinska Institute visualization core facility (KIVIF). The images were captured using the computer program QuantCapture 4.0 and the live and dead cells were identified and individually counted using the program QuantCount 3.0. Both programs were developed at the KIVIF using OpenLab Automator programming environment (Improvision). 15 control wells, that were used to determine the control cell survival (CCS), contained cells with only culture medium and 50 nl DMSO without drugs, 5 wells contained cells with culture medium alone. Comparing the two types of control wells no toxic effect

Comparison of the drug sensitivity pattern of the different LCLs
The 17  Remarkably, the LCLs of different origins showed a very similar sensitivity pattern against the different drugs. This phenomenon is illustrated in Figure 1, which summarizes the survival curves of all 17 LCLs against the four different concentrations of epirubicin. Each individual thin curve represents the average of three independent measurements. The thick line is the mean of all curves, the Mean Cell Survival (MCS). The grey shaded area marks the ± standard deviation (SD). Figure 2 summarizes the Mean Cell Survival curve along with the standard deviations for 28 drugs again emphasizing the very similar response of the different cell lines. The values for MG132 are represented only in Table 2.

Identification of highly effective and non-effective drugs
The starting concentrations of the dilution series (1×, 4×, 16×, 64×) for the individual drugs were initially determined based on the solubility of the different agents. To make the cell survival data comparable with each other, we first compiled the known Maximum Achieved Plasma Concentration data (Cmax) for all 28 drugs stated from the literature (Table 1- Table 2 along with MCS and ± SD values. Plotting the cell survival against the RMAPC values allowed a direct comparison of the effectiveness of different drugs. Figure 3 shows the comparison of the effectiveness of the members of three different drug families: the anthracyclines, vinca alcaloids and taxans. Figure 4 summarizes the mean cell survival curves for all the drugs along a common RMAPC axis. Based on their effectiveness on LCLs, we have divided the drugs into three different groups: Group 1: The drugs were considered to be highly effective if the mean cell survival (MCS) was below 30% at RMAPC 0.3 or below that.
Group 2: The drug was partially effective if MCS was under 60% and RMAPC was between 0.3 and 1.
Group 3: The drug was ineffective, e.g. the LCLs were resistant to the drug, if MCS was above 60% or RMAPC was above 1.
We found that the four most effective drugs against LCLs were: vincristine, paclitaxel, methotrexate and epirubicin (Group 1).
Although the Group 3 drugs were not effective against LCLs, these drugs show concentration dependent growth   inhibitory effects on other cell lines or primary tumors (data not shown).
No RMAPC could be determined for streptozocin because no pharmacokinetical trials with established Cmax plasma levels were found. RMAPC was also not available for MG132 because it has not been used clinically. Streptozocin had no effect on survival at the concentrations used in the study. Although at the highest concentrations the proteasome inhibitor MG132 effectively decreased the survival of the cells, its clinically licensed functional homologue, bortezomib, was not effective.
An alternative way to calculate relationship between the in vitro drug concentrations and the in vivo used values is to use the area under curve (AUC) values of the individual drugs. To compare AUC values with the RMAPC data the following formula was used: in vitro used concentration (µg/ml) × 72 h/AUC (µgxh/ ml) in vivo.
In vivo AUC levels from the literature are summarized in Table 1. Using the formula for representing the mean cell survival curves for all the drugs along a common "ratio of AUC" axis, carboplatin, methotrexate and dactinomycin showed higher efficacy in comparison to RMAPC figures whereas the sensitivity order for the other drugs or for the group classifications did not change.

Discussion
The presented data suggest that many EBV transformed Bcell lines share a common cytotoxic drug sensitivity profile independent of their origin. This profile does not change even after many years of in vitro culturing. EBV appears to be the necessary and sufficient etiological agent behind the malignant immunoblastic B-cell lymphoproliferations in immunosuppressed patients. In all these cases the EBV encoded, latency associated viral proteins drive the cell proliferation without any obvious need for additional genetic changes. The phenotype of these tumors closely resembles the in vitro growing LCLs and the experimentally induced tumors that appeared upon intraperitoneal implantation of LCLs into SCID mice [11]. Considering the phenotypic and karyotypic stability of EBV transformed B-cells, our data raises the hope that PTLDs, AIDS associated CNS lymphomas and XLP associated lymphoproliferations may show similar patterns of drug sensitivity as the ones that we have established in the present study on a cohort of diverse LCLs.
Sugimoto et al. suggested that EBV mediated transformation is a two step process where, after prolonged passage, the cells may become aneuploid, accumulate p53 mutations, down regulate the p16/retinoblastoma protein Drug sensitivity of LCLs for anthracyclines, vinca alkaloids and taxans Figure 3 Drug sensitivity of LCLs for anthracyclines, vinca alkaloids and taxans.
pathway and become tumorigenic upon subcutaneous inoculation into nude mice [7]. Some of the cell lines that we have examined were kept in continuous culture over five years (IARC171, LSspontan, 980215). These cells are still euploid and show low soft agar clonability. The oldest line IARC171, has wild type p53, is non-tumorigenic upon subcutan inoculation but produces generalized immunoblastic lymphomas in SCID mice after intraperitoneal inoculation. The findings that PTLD can arise very rapidly in post-transplant patient together with the data that freshly EBV infected B-cells can grow into generalized immunoblastic lymphomas within a few weeks, implies that the development of PTLD does not require aneuploidy. Sawada et al. reported that LCLs with negative or low telomerase activity and normal karyotypes are more sensitive against certain drugs, than LCLs with a high telomerase activity and abnormal karyotypes [22]. This data together with our present findings may suggest that rap-idly emerging PTLDs may show different drug sensitivity pattern from EBV positive aneuploid diffuse large cell Bcell lymphomas that arise for example in AIDS patients in sub-Saharan Africa.
Our data shows that euploid LCLs are particularly sensitive for anti-microtubule drugs and anthracyclines. Although all LCLs showed a dose dependent increase of cytotoxicity when treated with different members of these two drug families, only vincristine and paclitaxel as well as epirubicin and doxorubicin were considered to be effective when the used in vitro drug concentrations were compared to the maximum achievable plasma concentrations.
Alkylating agents, such as cyclophosphamide and ifosphamide were not effective on LCLs. This might be explained by the fact that both of these compounds are pro-drugs that have to be converted into active metabolites by the liver in vivo. Prednisolone was active but only in very high concentrations.

Conclusion
Vincristine and methotrexate are included in the frequently used CHOP and VAPEC-regimes, but no data could be found for the clinical use of epirubicin and paclitaxel for the treatment of PTLD or other EBV induced lymphoproliferations. Our results suggest that the inclusion of epirubicin and paclitaxel into chemotherapy protocols against PTLD may be justified.