Functional Bioassays for Immune Monitoring of High-Risk Neuroblastoma Patients Treated with ch14.18/CHO Anti-GD2 Antibody

Effective treatment of high-risk neuroblastoma (NB) remains a major challenge in pediatric oncology. Human/mouse chimeric monoclonal anti-GD2 antibody (mAb) ch14.18 is emerging as a treatment option to improve outcome. After establishing a production process in Chinese hamster ovary (CHO) cells, ch14.18/CHO was made available in Europe for clinical trials. Here, we describe validated functional bioassays for the purpose of immune monitoring of these trials and demonstrate GD2-specific immune effector functions of ch14.18/CHO in treated patients. Two calcein-based bioassays for complement-dependent- (CDC) and antibody-dependent cellular cytotoxicity (ADCC) were set up based on patient serum and immune cells tested against NB cells. For this purpose, we identified LA-N-1 NB cells as best suited within a panel of cell lines. Assay conditions were first established using serum and cells of healthy donors. We found an effector-to-target (E:T) cell ratio of 20∶1 for PBMC preparations as best suited for GD2-specific ADCC analysis. A simplified method of effector cell preparation by lysis of erythrocytes was evaluated revealing equivalent results at an E:T ratio of 40∶1. Optimal results for CDC were found with a serum dilution at 1∶8. For validation, both within-assay and inter-assay precision were determined and coefficients of variation (CV) were below 20%. Sample quality following storage at room temperature (RT) showed that sodium-heparin-anticoagulated blood and serum are stable for 48 h and 96 h, respectively. Application of these bioassays to blood samples of three selected high-risk NB patients treated with ch14.18/CHO (100 mg/m2) revealed GD2-specific increases in CDC (4.5–9.4 fold) and ADCC (4.6–6.0 fold) on day 8 compared to baseline, indicating assay applicability for the monitoring of multicenter clinical trials requiring sample shipment at RT for central lab analysis.


Introduction
Monoclonal antibodies targeting disialoganglioside GD 2 emerge as an important treatment option for NB, a dismal pediatric malignancy characterized by high expression of GD 2 on tumor cells [1,2]. Ganglioside GD 2 is a glycolipid antigen devoid of an intracellular signal transduction domain. Therefore the mechanism of action of anti-GD 2 monoclonal Ab mostly rely on immune effector functions mediated by mAbs, which are more and more recognized as the key features of this class of cancer therapeutics [3]. These features include the activation of CDC and ADCC.
CDC is induced through binding of a serine protease complex C1 to the Fc domains of two or more mAbs binding to antigens expressed on tumor cells. This classical complement pathway results in an activation cascade resulting in the membrane attack complex disrupting the target cell. ADCC is a result of Fc-gamma receptor (FccR) mediated interaction with effector immune cells such as natural killer (NK) cells, macrophages and granulocytes [3]. The binding of FccR to Fc domain induces both release of granzymes and perforin from effector cells leading to a target cell lysis and Fc-dependent tumor cell phagocytosis.
The clinical development of anti-GD 2 monoclonal antibodies for NB patients originated from the discovery of two distinct murine anti-GD 2 antibodies designated 3F8 [4] and 14.18 [5], respectively. High-risk NB patients were successfully treated within clinical trials with both antibodies mostly conducted by cooperating academic groups of pediatric oncologists. In a more multi center and international approach, the human/mouse chimeric version of 14.18 (ch14.18) has demonstrated activity and efficacy as a monotherapy [6,7] and in combination with cytokines [8]. In Europe, ch14.18 antibody was made available for clinical trials following the recloning of the antibody genes into CHO cells which was designated as ch14.18/CHO. This is important, as ch14.18/CHO revealed superior activity in mediating ADCC compared to ch14.18 antibody produced in other cell lines [9]. Subsequently, a validated industrial production process was established. This development was initiated by SIOPEN, a group of international clinical leaders in the field of neuroblastoma and funded by charities throughout Europe. Four European clinical trials with different treatment schedules of ch14.18/CHO are being conducted to investigate the influence of a combined immunotherapy of ch14.18/CHO, interleukin-2 (IL-2) and 13-cisretinoid acid on the outcome of patients with high-risk NB in the absence or presence of haploidentical blood stem cell transplantation. The first trial established the safety profile of ch14.18/ CHO in children with high risk NB [10]. The European phase III clinical trial (HR-NBL 1.5/ESIOP, Eudra CT: 2006-001489- 17) and the trial in the context of haploidentical stem cell transplantation (Eudra CT: 2009-015936-14) are based on a short term infusion of 20 mg/m 2 /d ch14.18 over 8 h on five subsequent days. To reduce side effects including neuropathic pain, a Phase I/II clinical trial was initiated based on the same cumulative dose of ch14.18/CHO (100 mg/m 2 /cycle) infused over a longer time period (ten days) (Eudra CT: 2009-018077-3). Within these trial protocols, a set of immune monitoring assays including the detection of ch14.18/CHO serum levels [11] and human anti-ch14.18/CHO immune responses [12], are implemented with the aim to identify immune biomarkers correlating with clinical response to ch14.18/CHO therapy. For a comprehensive assessment, validated bioassays to determine effector functions of ch14.18/CHO namely patient specific ADCC and CDC are of critical importance.
For analysis of patient-specific CDC and ADCC, we established and validated two non-radioactive and non-toxic cytotoxicity assays based on release of acetomethoxy derivate of calcein (calcein-AM), which is a membrane-permeable live-cell labeling dye. With these assays, we demonstrate GD 2 specific CDC and ADCC activity in ch14.18/CHO treated patients and demonstrated feasibility in the context of multicenter clinical trials.

Ethic statement
Participants were informed about the testing procedure and gave written informed consent. The study complies with the Declaration of Helsinki and was approved by the ethics committee of the University Medicine of Greifswald, Germany. De novo cell line establishment (HGW-1, HGW-2, HGW-3 and HGW-5 cell line) complies with the written informed consent from the donor, parents and legal guardians.
In order to establish new NB cell lines, tumor tissues and bone marrow samples from NB patients were used. Tumor tissue was first minced followed by homogenisation using 70 mM cell strainer (BD Biosciences, Heidelberg, Germany) and bone marrow samples were treated for 10 min with cold erythrocyte lysis buffer (0.15 M ammonium chloride, 10 mM potassium bicarbonate, 0.1 mM ethylenediaminetetraacetic acid, pH 7.4). After two wash steps (1x PBS, 5 min, 3006g, RT) cell pellets were resuspended in 10 ml 1x PBS and layered over 10 ml of Lymphocyte Separation Medium (PAA, Pasching, Austria). After centrifugation (30 min, 3006g, RT without brake) the upper layer was discarded, a layer of primary cells was carefully transferred into a new tube and the remaining Lymphocyte Separation Medium was washed off with 1x PBS supplemented with 0.5% BSA (Sigma Aldrich, Steinheim, Germany) and 2 mM EDTA (Sigma Aldrich, Steinheim, Germany) (pH 7.4, 10 ml, 5 min, 3006g, RT, brake on). Thereafter, the supernatant was carefully removed and the cell pellet was resuspended with 10 ml 1x PBS/0.5% BSA/2 mM EDTA buffer (pH 7.4) followed by cell counting and evaluation of cell viability. Finally, 1610 7 vital cells were used for isolation of GD 2 -positive cell subset with MACS technique (Miltenyi Biotec, Teterow, Germany).
For this purpose 1610 7 primary cells were resuspended in 200 ml 1x PBS/0.5% BSA/2 mM EDTA buffer (pH 7.4) and then incubated on ice with 20 ml FcR-blocking reagent (Miltenyi Biotec, Teterow, Germany) for 5 min. Thereafter, the murine anti-GD 2 Ab 14G2a (1.0 mg) was added followed by incubation for 10 min on ice. After washing with 1x PBS/0.5% BSA/2 mM EDTA buffer (15 ml, pH 7.4, 5 min, 3006g, +4uC) cell pellet was resuspended in 100 ml 1x PBS/0.5% BSA/2 mM EDTA buffer and 20 ml magnetic microbeads conjugated with anti-mouse IgG (Miltenyi Biotec, Teterow, Germany) were added followed by incubation on ice for 15 min. Next, the cells were washed again by adding 15 ml 1x PBS/0.5% BSA/2 mM EDTA buffer (pH 7.4) and centrifuged at 3006g for 5 min at +4uC. 1 ml of 1x PBS/ 0.5% BSA/2 mM EDTA buffer was used to resuspend the cell pellet and magnetic separation of GD 2 -positive cells was proceeded in MACS-separator (Miltenyi Biotec, Teterow, Germany) by applying the cell suspension onto the LS-separation column (Miltenyi Biotec, Teterow, Germany). The column was washed three times (3 ml, 1x PBS/0.5% BSA/2 mM EDTA buffer, pH 7.4) and then removed from the MACS-separator and placed into a collection tube. Cell fraction containing magnetically labeled GD 2 -positive cells were immediately flushed out by firmly applying the plunger supplied with the column (Miltenyi Biotec, Teterow, Germany). Finally, primary cells were washed with 1x PBS (pH 7.4, 5 min, 3006g, RT) and the cell pellet was resuspended in culture medium as described above.

Genetic characterization of de novo neuroblastoma cell lines
To determine DNA profiles of newly established cell lines, short tandem repeats (STR) analysis was done (Eurofins MWG Operon, Ebersberg, Germany). Evaluation of genetic aberrations of each newly established cell line was performed by high resolution SNP array analysis using the Affymetrix ultra HD SNP array (CytoScan HD Array). The combination of copy number with the allele information, i.e. zygosity status, enables the detection of loss of heterozygosity with copy losses as well as copy neutral allelic changes.

Analysis of neuroblastoma surface markers by flow cytometry
For analysis of GD 2 and CD56 surface expression, 1610 6 cells were stained with 1.0 mg ch14.18/CHO and 0.05 mg anti-CD56-APC (clone HCD56, BioLegend, Fell, Germany) in a total volume of 100 ml for 20 min, on ice, in the dark. The chimeric anti-CD20 antibody rituximab (Roche, Mannheim, Germany) and APC-labeled mouse IgG1 (clone MOPC-21, BioLegend, Fell, Germany) were used as negative controls for ch14.18 and anti-CD56, respectively. Cells were washed once with 1 ml wash buffer (1x PBS, 1% BSA, 0.1% NaN 3 , 0.1% EDTA, pH 7.4) (3006g, 5 min, RT). Supernatant was discarded and cells were incubated with 0.017 mg of PE-labeled anti-human IgG antibody (clone HP6017, BioLegend, Fell, Germany) in a total volume of 100 ml for 20 min, on ice, in the dark. Cells were washed once and resuspended in 500 ml wash buffer for flow cytometric analysis. To exclude dead cells from analysis, 4 ml of a 0.1 mg/ml DAPI solution (Sigma-Aldrich, Steinheim, Germany) were added 5 min prior to acquisition. For each sample 20,000 live cells were analyzed. Sample acquisition was performed at a BD FACS CANTOII using FACSDiva software (BD Biosciences, Heidelberg, Germany) and data was analyzed using FlowJo 7.6.1 software (Treestar, Ashland, OR, USA). For comparison of GD 2 expression on de novo NB cell lines, the ratio of mean fluorescence intensity (MFI) relative to isotype control was calculated according to the formula: geometric fluorescence mean of GD 2 stained sample/ geometric fluorescence mean of isotype control.

Tyrosine hydroxylase and MDR1 gene expression analysis by RT-PCR
Total RNA was isolated from #1610 7 cells using the RNA isolation Kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturer's instructions. RNA concentration was determined spectrophotometrically (BioPhotometer plus, Eppendorf, Hamburg, Germany) and 1.0 mg of total RNA was used for cDNA synthesis using qScript cDNA Synthesis Kit (Quanta BioSciences, Gaithersburg, MD, USA) according to the manufacturer's protocol.
Analysis of human TH (hTH) mRNA expression was performed using gene-specific primers designed with online primer design tool Primer3 and used for PCR. For the detection of hTH mRNA, the forward primer sequence used was 59-GGC CCA AGG TCC CCT GGT TC-39 and the reverse primer sequence was 59-ACA GCA GGC CGG CCA CAG GC-39. HTH was amplified by 25 cycles consisting of 95uC (30 s) for denaturation, 62uC (30 s) for primer-specific annealing, and 72uC (30 s) for elongation. The PCR product size was 405 bp. PCR without cDNA template (no template control) served as a negative control and GAPDH as a housekeeping gene (forward primer: 59-GAG TCA ACG GAT TTG GTC GT-39 and reverse primer: 59-TTG ATT TTG GAG GGA TCT CG-39, product size: 238 bp).
To evaluate expression of MDR1 gene, specific primers were designed as described above. Following primer sequences were used for PCR: 59-GCT CCT GAC TAT GCC AAA GC-39 (forward primer) and 59-TCT TCA CCT CCA GGC TCA GT-39 (reverse primer). MDR1 PCR product (202 bp) was amplified by 25 cycles consisting of 95uC (15 s) for denaturation, 61uC (15 s) for primer-specific annealing, and 72uC (15 s) for elongation. Negative control and GAPDH as an internal control (forward primer: 59-GAG TCA ACG GAT TTG GTC GT-39 and reverse primer: 59-TGT GGT CAT GAG TCC TTC CA-39, product size: 512 bp) were used as described above. After separation by electrophoresis on 2% agarose gel, PCR products were visualized by ethidium bromide (ImageJ Program). Expression of MDR1 gene product was analyzed relative to the internal control GAPDH.

Isolation of ganglidiomab from hybridoma supernatants
Ganglidiomab anti-idiotype (anti-Id) Ab was isolated from hybridoma cells as previously described [17]. Briefly, the concentration of ganglidiomab in hybridoma supernatants was determined by ELISA using ch14.18/CHO as a capture mAb [17]. Hybridoma supernatants containing 50 mg/ml of ganglidiomab were filtered and ganglidiomab was concentrated followed by binding to protein G as previously described. After wash steps ganglidiomab was eluted and quantified with ELISA using ch14.18/CHO as a capture mAb [17]. The final product contained 500 mg/ml of ganglidiomab and was used for further assay developments.

Serum preparation
Blood was collected using BD Vacutainer plastic serum tubes (BD Biosciences, Heidelberg, Germany). Serum was prepared from clotted blood samples by centrifugation for 10 min at 1,7006g, RT and stored in aliquots at 280uC. Serum used for ADCC assays was heat inactivated at 56u for 30 min. Serum as a source for complements in CDC assays was used without further manipulation.

Isolation of effector cells
Isolation of peripheral blood mononuclear cells (PBMCs). Blood was collected using BD Vacutainer plastic sodium-heparin tubes (BD Biosciences, Heidelberg, Germany) 6 ml of blood were carefully layered over 6 ml of Lymphocyte Separation Medium (PAA, Pasching, Austria). After centrifugation (30 min, 3006g, RT without brake) following four layers were defined: a) a clear upper layer (diluted plasma and platelets), b) a fluffy white layer of PBMCs, c) a clear Lymphocyte Separation Medium layer and d) a lowest substantial pellet of red blood cells containing granulocytes. After the upper layer was discarded, a layer of PBMCs was carefully transferred into a new tube and the remaining Lymphocyte Separation Medium was washed off twice with 1x PBS (10 ml, 5 min, 3006g, RT, brake on). Thereafter, the supernatant was carefully removed and the cell pellet was resuspended with 10 ml 1x PBS followed by counting of PBMCs.
To determine relative amounts of lymphocytes, granulocytes and monocytes in PBMCs, cells (2610 5 ) were resuspended in 2 ml 1x PBS and attached to glass slides using cytospin centrifugation (10 min, 2006g, RT). Cytospins were stained using the panoptic method of Pappenheim following the manufacturer's guidelines and a differential leucocyte count was performed by light microscopy. The lowest layer containing erythrocytes and granulocytes was used for further isolation of granulocytes.
Isolation of granulocytes. The lowest layer obtained during the isolation of PMBCs as described above was resuspended in 4 ml of cold erythrocyte lysis buffer (0,15 M ammonium chloride, 10 mM potassium bicarbonate, 0.1 mM ethylenediaminetetraacetic acid, pH 7.4) and incubated for 10 min on ice. Further, cell suspension was centrifuged for 5 min at 3006g (+4uC). Thereafter, the supernatant was carefully removed and the cell pellet was resuspended in 15 ml 1x PBS and centrifuged (5 min, 3006g, + 4uC). Then, granulocytes were resuspended in 10 ml 1x PBS and counted. The relative amount of granulocytes was determined by light microscopy following cytospin as described above.
Isolation of leukocytes. Leukocytes were isolated from 6 ml of sodium-heparin blood not older than 24 hours using cold erythrocyte lysis buffer. Briefly, 6 ml of whole blood was gently mixed with 12 ml of cold erythrocyte lysis buffer, incubated for 10 min on ice and centrifuged for 5 min at 3006g (+4uC). Thereafter, cell pellet was resuspended in 15 ml 1x PBS and centrifuged again (5 min, 3006g, +4uC). Finally, leukocytes were resuspended in 10 ml 1x PBS and counted.
Calcein-AM cytotoxicity assay. In order to evaluate patient-specific ch14.18/CHO-mediated CDC as well as ADCC, two calcein-AM-based release assays were established. For ADCC,  NB target cells labeled with membrane-permeable fluorescent dye calcein-AM were incubated with effector cells and serum of treated patients. For CDC, patient serum was incubated with calcein-AM-labeled target cells without effector cells.
Since high spontaneous release of calcein from labeled target cells is an obstacle for calculation of CDC, the level of spontaneous release was analyzed in a panel of GD 2 -positive NB cell lines (LA-N-1, CHLA-20, HGW-1, HGW-2, HGW-3 and HGW-5) and GD 2 -positive melanoma cell line M21. GD 2 -negative NB cell line SK-N-SH served as a negative control. 0.6610 6 cells were harvested, counted and washed twice with 10 ml 1x PBS (5 min, 3006g, RT). Cell pellet was then resuspended in 1 ml of respective culture medium and incubated with 10 mM calcein-AM (Sigma Aldrich, Steinheim, Germany) for 30 min at 37uC shaking at 100x rpm under CO 2 -free atmosphere (dark). After two wash steps in serum-free medium, supernatants were collected for background calculation. For spontaneous release determination, 5610 3 of calcein-AM labeled target cells were added to each well of a 96well plate (PAA, Pasching, Austria) and incubated with 12.5% heat-inactivated serum of a healthy donor (30 min, 56uC) for 4 h at 37uC, in the dark. To achieve final volume of 200 ml/well, serum-free cell culture medium (RPMI) was added. For maximum release, target cells were disrupted using ultrasonic homogenizer (30 s, RT). Then, supernatants (50 ml) of each well were transferred to a black 96-well plate for determination of fluorescence at 495 nm excitation and 515 nm emission wavelengths by Synergy HT multimode microplate reader (BioTek Germany, Bad Friedrichshall, Germany). Experiments were analyzed in triplicates using six replicate wells for spontaneous release (target cells only). Spontaneous release in percent was calculated according to the formula: (spontaneous releasebackground)/(maximum release -background)6100%.
Antibody-dependent cell-mediated cytotoxicity. A systematic analysis of the source of the effector cells and the effector to target (E:T) cell ratio was performed using 10 mg/ml of anti-GD 2 Ab ch14.18/CHO and LA-N-1 NB cell line as described above. Three effector cell populations (leukocytes, PBMCs and granulocytes) of a healthy donor (D1) were analyzed at different E:T ratios (80:1, 40:1 20:1, 10:1, 5:1, 2.5:1 and 1.25:1). Additionally, ADCC-mediated by activated effector cells was analyzed. Therefore, effector cells were cultivated for 64 h in RPMI culture medium supplemented with IL-2 (1000 IE/ml; Novartis, Nürnberg, Germany), 4.5 g/l glucose, 2 mM stable glutamine, 10% FCS and 1x P/S. After pre-incubation of labeled target cells (5610 3 /well) with ch14.18/CHO for 30 min at +37uC (dark), effector cells were added for further 4 h (+37uC, dark). Rituximab served as a negative control. GD 2 -specific ADCC of NB cells was demonstrated by pre-incubation of ch14.18/CHO with excess of anti-Id Ab ganglidiomab (50 mg/ml). To achieve a final volume of 200 ml/well, LA-N-1 cell culture medium was used. Finally, fluorescence levels of supernatants (50 ml) were measured as described above. Experiments were analyzed in triplicates using six replicate wells for spontaneous (target cells only) and three replicates for maximum release. ADCC was calculated according to the formula: (test release -spontaneous release)/(maximum release -spontaneous release)6100%.
To investigate the sensitivity, a concentration range of ch14.18/ CHO mAb inducing measurable ADCC, serial Ab dilutions were prepared (10,000, 1,000, 100, 10, 1, 0.1 and 0.01 ng/ml). ADCC evaluation was performed using LA-N-1 as a target cell line and activated leukocytes of a healthy donor (D1) as effector cells at E:T ratio of 40:1 as described above. Rituximab and ganglidiomab containing controls were included as described above.
Assay validation and run acceptance analysis. To show reproducibility, accuracy and precision of the established cytotoxicity assay, both within-assay and inter-assay precision analysis were determined according to international guidelines for bioanalytical method validation [18]. For both data validation and run acceptance analysis, we included in each analytical run a quality control containing a known concentration of ch14.18/ CHO mAb (1.0 mg/ml) prepared in serum of a healthy donor (D1).
Within-assay precision was determined with ten triplicated samples containing 1.0 mg/ml ch14.18/CHO in serum of a healthy donor. The within-assay precision value was calculated as follows: (standard deviation (SD) of replicates divided by mean of replicates)6100%. To confirm consistency of measurements over time, the inter-assay precision was calculated. For this, ten triplicated samples containing a defined concentration of ch14.18/CHO (1.0 mg/ml) prepared in serum of a healthy donor were analyzed by different operators on different days. The interassay precision was calculated according to the formula: (SD of replicates divided by mean of replicates)6100%.
For validation of cytotoxicity analysis, we developed a set of tailored QCs containing defined concentrations of ch14.18/CHO (1.0 mg/ml) prepared in serum of a healthy donor. For each serum batch the level of cytotoxicity was first determined in at least six independent measurements by different operators on different days. Then, we calculated the CV for these assays and plotted the results over time. Only the serum batch spiked with ch14.18/ CHO showing CV under 20% was used as QC and included in each analytical run for run acceptance analysis.
To accept or reject analytical runs, QCs were included in each run. Deviation of the QC results from their nominal value was calculated as absolute value (modulus) using the formula: | 100% -(MEAN/nominal value)6100% |, according to international guidelines for bioanalytical method validation [18]. If the differences between the QC and its respective nominal values exceeded 20%, the run was rejected and the data were not included in the data analysis. Only runs including QC deviating from the nominal within 20% were accepted.
Stability of CDC and effector cell viability. To evaluate the impact of shipping condition on CDC, viability and cell count of effector cells, we analyzed both serum and whole blood samples of a healthy donor (D1) stored under different conditions or subjected to five freeze-thaw cycles. The serum samples were prepared as described above. For evaluation of repeated freezing and thawing or storing at RT (up to 96 h), serum samples aliquots containing two defined ch14.18/CHO concentrations (1 and 0.1 mg/ml) were used. For evaluation of leukocyte viability and cell count, whole blood samples were collected in either BD Vacutainer plastic K2EDTA-(BD Biosciences, Heidelberg, Germany) or BD Vacutainer plastic sodium-heparin-containing tubes (BD Biosciences, Heidelberg, Germany) and stored at RT for up to 72 h. CDC was evaluated using the established cytotoxicity assay. Rituximab containing samples and Ab-free sera were used as negative controls. Viability of leukocytes was evaluated using trypan blue test and a cell count with automated cell counter (EMD Millipore Corporation, Billerica, MA, USA).
Statistics. After testing for normality and equal variance across groups, differences between groups were assessed with the Students t-test. A P level of , 0.05 was considered significant. All data are given as means 6 SD or means 6 SEM. Analysis was performed using the software SigmaStat (Jandel, San Rafael, CA).  (Table 1). HGW-3 and HGW-5 cells also showed the typical deletion of chromosome 1p. HGW-1 showed gain of 11q in contrast to the other lines ( Table 1). Analysis of HGW-2 is pending. Finally, cell line DNA profiles determined with STR analysis clearly showed the de novo nature of each cell line.

Target cell line selection to establish CDC and ADCC bioassays
Further characterization of these new HGW cell lines were done in comparison to NB cell lines LA-N-1, CHLA-20 and SK-N-SH as well as a melanoma cell line M21.
Since GD 2 is the target antigen of ch14.18, its expression on the target cell line is crucial to set up CDC and ADCC bioassays. The GD 2 expression analysis was complemented with CD56 since both surface molecules are typical features of NB cells.
NB origin of newly established cell lines HGW-1, HGW-2, HGW-3 and HGW-5 was confirmed by detection ob both CD56 and GD 2 NB expression markers (Fig. 1B). All de novo cell lines showed homogeneous expression of CD56 and heterogeneous expression of GD 2 (Fig. 1B). To compare GD 2 expression profiles, MFI ratios were used. The highest MFI ratio was found on HGW-1 (63.4). HGW-2 and -3 displayed MFI ratios of 46.9 and 22.9, respectively. The lowest MFI ratio was calculated for HGW-5 (5.0). Interestingly, GD 2 expression analysis of HGW-5 revealed two distinct subpopulations. About 84% of cells had a GD 2 dim phenotype (MFI ratio 3.0) and the smaller subpopulation of about 16% had a GD 2 bright expression pattern (MFI ratio 80.5; Fig. 1B). In contrast, HGW-3 cell line which consists of two morphologically distinctive cell populations (adherent and suspension cells) did not show differences in GD 2 expression profiles.
The NB cell lines LA-N-1 and CHLA-20 used as controls revealed high expression of both NB marker GD 2 and CD56 (Fig. 1A). In contrast, SK-N-SH showed high expression of CD56 but lacked expression of GD 2 (Fig. 1A), therefore, SK-N-SH was excluded from consideration as a target cell line for the bioassays. As expected, a melanoma cell line M21 did not show CD56 expression but a high expression of GD 2 (Fig. 1A).
In order to further characterize the newly established HGW cell lines, we also investigated expression of hTH, the first step enzyme of catecholamine biosynthesis, which is a typical feature of NB cells. Expression of hTH mRNA was found in all newly established NB cell lines HGW-1, HGW-2, HGW-3 and HGW-5 as well as in NB cell lines LA-N-1 and CHLA-20 ( Fig. 2A) in contrast to the melanoma cell line M21 used as a negative control. The NB cell line SK-N-SH was surprisingly hTH negative further supporting the decision to exclude this cell line from further consideration.
The analysis of MDR1 gene expression was used as further selection criterion since calcein is an MDR1 substrate. Therefore, high MDR1 expression results in active transport of calcein into the extracellular space which accounts for a high background in calcein release assays [19]. The weakest mRNA expression of MDR1 was found in LA-N-1. (Fig. 2B-C). CHLA-20 and HGW-1 showed about three-fold and SK-N-SH, M21, HGW-2 and -3 about two-fold higher levels of MDR1 mRNA expression compared to LA-N-1 (Fig. 2B-C). These findings correlate with spontaneous release data for cell lines analyzed revealing the lowest level of spontaneous release in LA-N-1 cells (Fig. 2D).

Assessment of variables for the complement-dependent cytotoxicity bioassay
The optimal serum dilution used for the CDC bioassay with patient samples was evaluated by using a serial 1:2.dilution (undiluted, 1:2, 1:4, 1:8, 1:16, 1:32, 1:64, 1:128, 1:256) of serum of a healthy donor in the presence and absence of two distinct concentrations of anti-GD 2 Ab ch14.18/CHO, i.e. 1.0 and 0.1 mg/ml, respectively. Interestingly, we found a bell-shaped curve of ch14.18-mediated CDC activity with decreasing serum concentrations at both ch14.18 concentration levels (Fig. 3A).  . Within-assay precision, inter-assay precision and sample stability. For reliable and reproducible evaluation of cytotoxicity, withinassay and inter-assay precision analyses were performed (A and C). Both parameters were calculated according to the formula: SD/mean6100% and found to be under 20%. To determine within-assay precision (A), triplicated serum samples of a healthy donor (12.5% final concentration) with defined ch14.18/CHO concentrations (1.0 mg/ml) and calcein-labeled LA-N-1 target cells were analyzed. The cytotoxicity analysis was repeated ten times on the same plate. Results are presented as mean CDC of triplicates 6 SD for ten data sets. Inter-assay precision (C) was determined on different days by different operators. Ten independent measurements of serum samples containing 1.0 mg/ml ch14.18/CHO and calcein-labeled LA-N-1 target cells were performed. The analyzed cytotoxicity levels are given as mean values 6 SD for ten independent assays. To determine stability of CDC (B and D), two ch14.18/CHO concentrations 1.0 mg/ml (closed circles) and 0.1 mg/ml (open circles) prepared in 12.5% serum of a healthy donor were analyzed with established cytotoxicity assay. Samples were subjected to either storage at room temperature for up to 96 h (B) or to five freezethaw cycles (D). Rituximab containing (closed triangles) and Ab-free controls (open triangles) were included as described in Materials and Methods. Data are shown as mean CDC values 6 SE of two independent experiments performed at least in triplicates. doi:10.1371/journal.pone.0107692.g008  3A, white columns). Again, the undiluted as well as 1:128 and 1:256 diluted sera (final concentrations: 3.1% and 1.6%, respectively) could not induce measurable CDC-mediated target cell lysis, and CDC-mediated by 1:2 diluted serum was found to be only at 14%. In summary, the highest CDC-mediated target cell lysis was observed by both concentrations of ch14.18/CHO (1.0 and 0.1 mg/ml) to be between 1:4 and 1:32. Based on these data we defined the serum dilution at 1:8 as the standard procedure to evaluate CDC in treated patients.
To further confirm that the cytotoxic activity observed in these experiments is mediated by the complement system, we added the complement inhibitor eculizumab (trade name Soliris) to our system (Fig. 3B, closed triangles). Ch14.18 (1.0 mg/ml) prepared in serum of a healthy donor (D1) (12.5% final concentration) induced a LA-N-1 target cell lysis of 100%. Co-incubation with a serial dilution of 1.0 mg/ml eculizumab inhibited the cytotoxic activity in a concentration-dependent manner resulting in almost complete abrogation of ch14.18-mediated target cell lysis (Fig. 3B, closed triangles). These findings prove the complement-mediated cytotoxic effect. GD 2 specificity was also demonstrated by preincubation of ch14.18 with excess of anti-Id Ab ganglidiomab (Fig. 3B, open circle) or by replacing ch14.18 with similar concentration of mAb rituximab used as a negative control (Fig. 3B, closed square).
In order to make a final decision on the target NB cell line to be used for the bioassays, we analyzed CDC against our panel of cell lines with 12.5% serum (1:8 dilution) as a source of complement and 1.0 mg/ml ch14.18/CHO (Fig. 3C). The highest level of ch14.18/CHO-mediated CDC was observed using LA-N-1 (97%) NB cells compared to all other cell lines analyzed in particular HGW-5. These differences in GD 2 -specific lysis correlate with GD 2 expression profiles of NB cell lines analyzed with flow cytometry (Fig. 1B). GD 2 negative SK-N-SH cells were used as a negative control. Importantly, pre-incubation with excess of anti-Id ganglidiomab resulted in complete abrogation of CDC in that cell line. Interestingly, for some cell lines, we found antibodyindependent target cell lysis in samples incubated with 1.0 mg/ml non-specific rituximab or excess of anti-Id ganglidiomab (CHLA-20, HGW-1, HGW-5 and M21), indicating the activation of Abindependent alternative complement activation pathways. In summary, LA-N-1 showed the highest sensitivity and specificity for GD 2 -specific CDC-mediated by ch14.18/CHO which was therefore selected as the designated cell line to establish the bioassays.
The impact of the source of complement on ch14.18/CHOmediated CDC against LA-N-1 NB target cells was analyzed using sera of four healthy donors (D1-4) (Fig. 4) and compared to that of rituximab used as a negative control. The cytotoxic activity mediated by 1.0 mg/ml ch14.18/CHO did not reveal significant inter-individual variations with a maximum activity between 90-100%. GD 2 specificity was demonstrated by a complete inhibition of CDC-mediated cytotoxicity to the level of the negative control (rituximab) after pre-incubation of serum samples with 5.0 mg/ml anti-Id Ab ganglidiomab. Interestingly, different inter-individual levels of ch14.18/CHO-independent anti-tumor cytotoxicity could be observed in serum samples of the selected donors in a range of 10% to 40%, indicating the presence of natural CDC activity against LA-N-1 as a result of complement activation mediated by the non-classical pathway [20]. Ch14.18/CHO concentrations below 10 ng/ml did not mediate detectable CDC in all donor sera analyzed. Based on these data we used for further experiments serum of a donor D1 lacking Ab-independent cytotoxicity (Fig. 4A).
CDC-mediated by different anti-GD 2 mAb constructs including 14G2a, ch14.18/CHO, hu14.18-IL-2 was compared to complement fixation deficient mutants hu14.18 K322 [21] and ch14.18delta-CH2 against LA-N-1 NB target cells using serum of a healthy donor D1 described above (Fig. 5). The level of a concentration-dependent CDC activity observed was highest with ch14.18/CHO followed by 14G2a and hu14.18-IL2 (Fig. 1B and D) compared to rituximab negative controls. The activity was GD 2 -specific since pre-incubation of samples with 5.0 mg/ml anti-Id ganglidiomab completely inhibited target cell lysis (Fig. 5). As expected, no detectable CDC could be observed with anti-GD 2 antibody constructs which lack complement binding sites (ch14.18-delta-CH2 and hu14.18 K322) ( Fig. 5C and E). In summary, the CDC assay revealed consistent GD 2 -specific results related to the presence and absence of a complement fixation moiety within the anti-GD 2 construct, with the highest level for ch14.18/CHO.

Antibody-dependent cell-mediated cytotoxicity bioassay
The analysis of ADCC was also performed using LA-N-1 NB target cells labeled with calcein as described in Materials and Methods. Assay conditions related to effector cell preparation and determination of an optimal effector-to-target cell ratio were done with blood samples of a healthy donor D1 in the presence of 10 mg/ml ch14.18/CHO compared to rituximab and to samples incubated with ch14.18/CHO and anti-Id used as negative controls (Fig. 6A-I) as well as patients treated with ch14.18/ CHO (Fig. 6J), respectively.
For the analysis of ch14.18/CHO-mediated ADCC, three distinct sources of effector cells were systematically evaluated. The first source consisted of a leukocyte preparation following red blood cell lysis of anti-coagulated blood (Fig. 6C). The second and third source was prepared from blood samples centrifuged in a blood separation media as described in Materials and Methods. Lymphocyte rich PBMCs were collected from the fluffy white layer (Fig. 6B) and a granulocyte rich cell fraction was collected from the lowest layer followed by lysis of erythrocytes as described above (Fig. 6A). Additionally, amounts of lymphocytes, granulocytes and monocytes were determined in both PBMC and granulocyte cell fractions showing about 80% lymphocytes, 10% granulocytes and 10% monocytes and 25% lymphocytes, 72% granulocytes and 3% monocytes in PBMC and granulocyte rich fraction, respectively.
The initial evaluation included such effector cells prepared from a healthy donor D1 either used immediately after isolation (0 h; Fig. 6D-F) or after incubation in medium containing 1,000 IU/ml IL-2 for 64 h (Fig. 6G-I). The IL-2 incubation was done in order to simulate the use of IL-2 in combination with ch14.18/CHO in clinical trials.
Of the fresh isolated effector cell preparations used immediately after isolation, only the granulocyte rich cell fraction mediated a significant ADCC reaction at an E:T ratio of 80:1. This effect was GD 2 -specific since anti-Id ganglidiomab completely inhibited ch14.18-mediated ADCC (Fig. 6F). In contrast, both freshly isolated leukocytes and lymphocyte rich PBMCs did not show significant cytotoxic activity against LA-N-1 NB target cells at any E:T ratio (Fig. 6D-E). This sharply contrasts to the response observed following IL-2 incubation (Fig. 6G-H). A strong increase of ADCC-mediated by leukocytes and lymphocyte rich PBMCs (Fig. 6G-H) but not by the granulocyte rich fraction was observed (Fig. 6I). All responses were GD 2 -specific and correlated with the E:T ratio. In conclusion, leukocytes prepared by red blood cell lysis of anti-coagulated blood and the lymphocyte rich PBMC preparations revealed equivocal ADCC responses at an E:T ratio of 20:1 and 40:1.
In order to confirm these results, we isolated leukocytes and lymphocyte rich PBMC preparations from the same blood sample obtained from five NB patients treated with a combination of IL-2 and ch14.18/CHO (Fig. 6J). Again, both effector cell preparations revealed inter-exchangeable results when used at an E:T ratio of 20:1 (PBMCs) and 40:1 (leukocytes) in contrast to rituximab controls, respectively, which in conclusion are both appropriate for immune monitoring.
The sensitivity of the ADCC reaction was determined using leukocytes prepared by red blood cell lysis of anticoagulated blood of a healthy donor (D1) cultivated in medium containing 1,000 IU/ml IL-2 for 64 hours at an E:T ratio of 40:1 in the presence of varying concentrations of ch14.18/CHO (Fig. 7). A concentration-dependent ADCC was observed with a limit of detection at 10 ng/ml ch14.18/CHO. Again, the response was GD 2 -specific, since pre-incubation with anti-Id ganglidiomab inhibited the cytotoxic activity to the level of background responses observed with rituximab used as a negative control.

Validation of bioassays, quality controls, run acceptance criteria and sample stability
The within-assay precision and the inter-assay variation were assessed according to international consensus recommendations for the bioanalytical method validation [18]. We defined values of CV for both within-and inter-assay variation up to 20% as acceptable. For within-assay precision, CDC of ten samples containing 1.0 mg/ml ch14.18/CHO prepared in serum of a healthy donor (D1) was performed and CDC data were used for calculation of CV (Fig. 8A). Each sample was analyzed in duplicates on the same plate. Analysis revealed a CV of 9% indicating reproducible performance of the assay. The inter-assay CV was determined in triplicates by changing the operators on ten different days. This analysis revealed a CV of 12% (Fig. 8C) clearly showing that results obtained are operator-independent and consistent over time.
To accept or reject experimental runs, QCs containing a known concentration of ch14.18/CHO (1.0 mg/ml) in serum of a healthy donor (D1) were included in each analysis. The deviation coefficients were calculated using QC results obtained after each analysis and the respective QC nominal value. The differences did not exceed 20% allowing analysis acceptance.
In order to test sample stability and subsequent use in the bioassays described after distinct storage conditions, we used healthy donor serum (D1) containing two defined concentrations of ch14.18/CHO (1.0 and 0.1 mg/ml). Neither storage at RT for up to 96 hours (Fig. 8B) nor up to three repeated freeze thaw cycles (Fig. 8D) revealed significant changes in the level of CDCmediated by ch14.18/CHO. Next, we analyzed leukocyte count and viability in blood samples using two anticoagulants, EDTA and sodium-heparin, respectively (Table 2). In general, sodiumheparin anticoagulated whole blood revealed longer stability compared to EDTA anti-coagulated blood with acceptable quality for up to 48 h storage at RT. Blood stored at RT for 72 h negatively affected both leukocyte count and viability independent on anticoagulants. These data were confirmed by ADCC analysis using effector cells prepared from samples stored as described above.
We also evaluated the impact of storage conditions on the granulocyte rich effector cell fractions. We prepared them from blood samples after 2 and 4 hours of storing at RT and included granulocyte rich fractions freshly isolated, but cultured for 64 h as described in Materials and Methods (data not shown). We could observe a reduction in granulocyte count by only 17% and 21% at 2 and 4 hour time points, respectively. Interestingly, the cell viability of granulocytes rich fractions was 99% at both time points. The cell count of granulocytes was reduced by 60% after 64 h of culture with nearly 100% viability consistent with the known fragility of this cell population. This finding is in line with a reduction of the cytotoxic efficacy by 50% compared to fresh isolated granulocytes at the 64 h time point. Therefore, analysis of granulocyte mediated ADCC clearly requires fresh blood samples and timely processing.
In summary, sodium-heparin anti-coagulated blood and serum can be stored together at RT for up to 48 h allowing for a timely transfer of the samples from clinical study sites for central lab analysis of CDC and ADCC using the described bioassays.

Evaluation of CDC and ADCC bioassays in neuroblastoma patients
Serum samples and anti-coagulated blood (heparin) were collected from three selected NB patients (patient 1, patient 2, patient 3) treated within a Phase I/II clinical trial where patients are treated with a cumulative dose of ch14.18/CHO (100 mg/ m 2 /cycle) infused over a time period of ten days (Eudra CT: 2009-018077-3). The collected samples were taken at base line before start of immunotherapy and on day 8 after start of antibody infusion. The collected serum samples were used after heatinactivation for ADCC-and untreated for CDC-assays containing a patient-specific level of ch14.18/CHO Ab as a result from the treatment (i.e. without exogenous addition of ch14.18/CHO). The serum levels of ch14.18/CHO were determined as previously described [11] (P1: 13.3060.08 mg/ml, P2: 15.1460.24 mg/ml, P3: 11.3960.24 mg/ml).
The effector cells were also used without further ex vivo stimulation at an E:T ratio of 40:1.
ADCC and CDC analyzed with these blood and serum samples resulted in a significant increase of CDC (Fig. 9A) and ADCC (Fig. 9B) in all treated patients. When serum samples were replaced with rituximab or pre-incubated with excess of anti-Id Ab ganglidiomab, only background cytotoxicity was observed in both bioassays. In summary, the ADCC and CDC methodology described here indicate GD 2 -specific cytotoxic activity in blood and serum samples obtained from patients treated with ch14.18.

Discussion
Ab are important therapeutics for the treatment of malignant disease and many mechanisms have been proposed to explain the clinical anti-tumor activity of mAb targeting a tumor associated antigen. Some Ab act through their ability to disrupt signaling pathways involved in the maintenance of their malignant phenotype such as blockade of ligands binding to growth factor receptors (e.g.cetuximab; EGFR) [22]. However, the ability of Ab to initiate a tumor-specific immune response has been less well recognized. Several studies have established the importance of ADCC-mediated through Fc-Fcc receptor interaction as well as CDC for the in vivo anti-tumor effects of mAb. This was demonstrated in mice which lack Fcc receptor expression or which are deficient in C1q, a critical component of the complement system. In both types of mouse strains the antitumoral activity of anti-CD20 Ab was diminished compared to wild-type strains [23,24].
The initiation of a tumor-specific immune response is also an important mechanism of action related to passive immunotherapies based on application of anti-GD 2 mAb. This intervention emerges as an established concept for the treatment of NB, a challenging malignancy in pediatric oncology [25]. The target of this strategy is ganglioside GD 2 , a glycolipid expressed on the cell surface of malignant cells of neuroectodermal origin devoid of an intracellular signaling domain. Therefore, the evaluation of tumorspecific anti-NB immune responses mediated by ADCC and CDC should capture the core components of the clinical activity observed with this class of anti-tumor Ab.
Here, we describe two rapid non-radioactive and non-toxic ADCC and CDC assays allowing for analysis of samples collected from NB patients treated with anti-GD 2 Ab ch14.18/CHO for the purpose of immune monitoring of ongoing clinical trials. Importantly, we demonstrated increased GD 2 specific CDC and ADCC in three selected high-risk NB patients treated with ch14.18/CHO indicating expected effector functions of ch14.18/ CHO in treated patients.
Among non-radioactive agents that can be used for such cytotoxicity assays, calcein-AM is one of the most prominent fluorescent dyes [26] used for labeling of target cells. Here, expression of multidrug transporter P-glycoprotein (MDR-1) is a hindering factor extruding calcein from target cells independently of CDC or ADCC. To avoid this problem, we selected LA-N-1 cells from a panel of NB cell lines. LA-N-1 showed low Pglycoprotein expression levels and subsequently low background signals after calcein labeling. Moreover, the most homogeneous and strongest GD 2 expression and the highest level of GD 2 -specific lysis in contrast to all other NB cell lines of the available panel resulted in the decision to establish CDC and ADCC bioassays with LA-N-1 target cells. We also aimed to optimize the bioassays to reduce both preparation time required for effector cell isolation as well as sample volume, which is particularly important in a pediatric population. To this end, a minimum of 100 ml of serum and a blood sample containing 200,000 leukocytes are sufficient which can be isolated within 10 min from small blood volume of less than 500 ml.
A second challenge to overcome is the fact that European NB trials are generally multi-center in nature and require sample transfer to a central laboratory for analysis. Therefore, the impact of shipping conditions is of great importance. According to our present data, serum samples could be stored at RT for four days and subjected three times to repeated thawing and freezing without significant influence on CDC. For the analysis of ADCC, blood needs to be anticoagulated. Although EDTA is the most commonly used anticoagulant, it is known that it affects both erythrocytes and leukocytes, causing membrane damage [27]. Here, we compared the impact of two anti-coagulants, EDTA and sodium-heparin, on leukocyte count and viability and count in blood samples stored at RT for 72 h. We could not observe any significant changes in cell-viability and -count of effector cells in sodium-heparin-anticoagulated blood for up to 48 h at RT in contrast to EDTA-anticoagulated blood samples. In contrast, the analysis of granulocyte rich effector cell fractions revealed a reduction of granulocyte counts and anti-tumor cytotoxic activity during storage of blood samples. Therefore, the evaluation of granulocyte-mediated ADCC as a biomarker requires timely processing and analysis, which is unsuitable for central lab testing in a multicenter setup of a clinical trial. In summary, serum and in sodium-heparin-anticoagulated blood samples can be shipped within 48 h for central laboratory bioassay analysis. We analyzed CDC and ADCC in three selected high-risk NB patients participating in this trial and could observe a significant increase of both anti-tumor CDC and ADCC after Ab administration indicating beneficial effects of this immunotherapy.
In order to permit a valid interpretation of data from functional assays employed for quantitative determination of cytotoxicity in biological samples, reproducible and reliable data must be generated operator independent in a laboratory performing such analyses [28]. To validate the established assay and to show its reproducibility, accuracy and precision, both within-assay and inter-assay precision analysis were determined and QC were included in each analytical run according to the international guidelines for bioanalytical method validation [18]. We found CV of both within-assay and inter-assay to be under 20% indicating operator-independent and reproducible performance over time.
Recently, a concept of a personalized medicine in NB treatment becomes more important [29]. In line with the fact that the extraordinary diversity of the clinical course in NB patients is reflected by a substantial genetic variation [29], we observed interindividual differences in a basic Ab-independent cytotoxicity in NB patients as well as about 40% inter-individual differences in healthy donors anylzed. By using the bioassays described here in combination with pharmacokinetic [11] and human anti-chimeric response analysis [12], we aim to monitor patients currently treated in clinical trials with ch14.18/CHO in order to work towards the establishment of a set of biomarkers potentially predictive of response.
Taken together, we reported two validated non-toxic and nonradioactive in vitro assays for rapid analysis of patient-specific CDC and ADCC in small volume samples collected from NB patients treated with anti-GD 2 Ab. These two assays allow reliable analysis within 48 h after sample collection for standard laboratory purposes and large-scale applications in clinical trials and are therefore a suitable tool for immune monitoring also in the multicenter setting.