1

Supplementary Information

Patient Inclusion/Exclusion Criteria

Eligible patients had histologically confirmed EBV-positive malignancy and were either in first remission or had persisting disease for which there was no standard therapy. All patients were in complete (or unconfirmed complete) remission, at least 12 weeks post-completion of first-line treatment (either as definitive chemoradiotherapy or palliative chemotherapy) and met the following inclusion criteria: toxicities resolved to ≤grade 1, ≥18 years, using adequate birth control, performance status 0-1, life expectancy >4 months, normal hepatic function (alkaline phosphatase, alanine amino-transferase (ALT) and/or aspartate amino-transferase (AST) < 1.5 x ULN), normal renal function (calculated creatinine clearance > 50ml/min), haemoglobin (Hb) > 10.0 g/dl; Neutrophils ≥ 1.5 x 109/L; Platelets (Plts) ≥ 100 x 109/L. The lymphocyte count had to either exceed the lower limit of normal in the investigator site or exceed 0.5x109/L with no successive rises on three occasions >3 weeks apart. Patients were excluded if they were known to be Hepatitis B, Hepatitis C or Human Immunodeficiency Virus (HIV) positive, had current active autoimmune disease or skin diseases requiring therapy, on-going active infection, a previous severe allergy to egg products, previous myeloablative therapy, splenic dysfunction or were undergoing immunosuppressive therapy.

Vaccine

The MVA-EL vaccine was supplied as a frozen aqueous suspension of 8.51 x 108 pfu/ml in 10 mM Tris buffer containing 140 mM NaCl, pH 7.7, manufactured by Impfstoffwerk Dessau-Tornau (IDT). The same batch of vaccine was used in the Phase 1 trial described herein and the previously reported Phase 1 trial conducted in Hong Kong.

Blood Sampling

Whole blood samples (typically 40ml) were collected into lithium heparin vacutainers during screening, on day 1 and 8 of each cycle and at week 10, week 11, week 14, month 6 and month 12. Blood was diluted with an equal amount of RPMI medium, layered onto a cushion of Ficoll density medium and centrifuged according to manufacturer’s recommendations. Aliquots of cell-free plasma were collected before harvesting peripheral blood mononuclear cells (PBMC) from the top of the ficoll cushion. Plasma samples were frozen and PBMCs cryo-preserved by controlled cooling in medium containing fetal calf serum and DMSO prior to long-term storage in the vapor phase of liquid nitrogen. Samples were transported in liquid nitrogen to the School for Cancer Sciences, University of Birmingham for analysis. An additional 4ml EDTA whole blood sample was also collected for HLA class I and II typing by the National Blood Service, Vincent Drive, Edgbaston, Birmingham. HLA A2 subtyping was performed by Dr Martin Barnardo, Transplant Immunology, Oxford Transplant Centre, Churchill Hospital, Oxford, UK.

EBV Genome Levels

EBV genome levels in plasma were measured by real time quantitative PCR as described (Gallagher et al., Int J Cancer 84, 442-448, 1999) in the Health Protection Agency Laboratory, Birmingham Heartlands Hospital, UK

2

EBNA1 Antibody Levels

EBNA1 specific IgG antibodies were measured using a commercial diagnostic ELISA (Diamedix, Miami FL) modifying the manufacturer’s protocol to allow changes in antibody levels to be detected. Thus, an initial experiment was first performed in which a dilution series of each patient’s pre-vaccination plasma sample was tested in the ELISA. This identified, for each patient, a plasma dilution that elicited an ELISA absorbance value in the midpoint of the assay’s linear range. All plasma sample from a single patient were then diluted the same and tested in the ELISA – since the pre-vaccine sample is in the middle of the linear range this allowed any increase or decrease in antibody levels to be detected.

ELIspot Assays

ELIspot assays (Mabtech, Stockholm, Sweden) were performed to detect IFNγ-secreting epitope-specific T cells. For each patient, samples taken from across the whole vaccine course were thawed, rested overnight and then tested in the same assay. Cell input was 3x105 cells/well in all assays. In most assays cell numbers permitted the use of mean readings from two or three identical wells. Patient’s cells from each time point were always tested with an appropriate negative control as well as the T cell mitogen phytohaemagglutin to confirm the cells were functional in the assay. Standardized controls included in the assays were PBMC from healthy volunteers (buffy coats from the National Blood Service, Birmingham) harvested in large numbers and cryopreserved in multiple small aliquots that had known performance properties in ELIspot assays. After assay completion, individual cytokine-producing cells visualized as dark spots (AP color development kit, Biorad, London, UK) were counted using an automated reader (AID GmbH, Strasberg, Germany). ELIspot data was accepted provided there was complete documentation of the procedure, noting any deviations. Recorded operator error would have resulted in affected wells being excluded from analysis. Results of the standardized control samples confirmed assays had been undertaken correctly.

Two types of ELIspot assays were undertaken for the study. The first used Pepmixes, pools of 15-mer peptides overlapping by 11 residues that span an antigen’s entire primary sequence (JPT Peptide Technologies GmbH, Berlin, Germany). The Pepmixes used included EBNA1 and LMP2 as well as the non-vaccine sequences EBNA 3A and combined influenza matrix and nucleoprotein (termed “FLU”); actin Pepmix served as the negative control. PBMC from three time-points (pre-vaccination, post cycle 2 and post cycle3) were tested. For each PBMC sample, the readings for antigen recognition were adjusted by subtracting the readings of background IFN-γ release in control wells tested against actin peptides.

The second type of ELIspot assay used peptides corresponding to defined HLA I- and HLA II-restricted T cell epitopes (Figure 1 and listed in Supplementary Table 1) to analyze the T cell response in finer detail. Peptides (synthesized by Alta Bioscience, Birmingham UK) were selected for use in each assay on the basis of the patient’s HLA type and were used in the assay individually or, in cases where cell numbers were limiting, in pools of two or three peptides (when used in pools the concentration of each peptide was maintained). DMSO solvent, added to wells at the same concentration, served as the negative control. For each PBMC sample, the readings for peptide recognition were adjusted by subtracting the readings of background IFN-γ release in control wells tested against DMSO.

PBMC from screening, day 1 and 8 on each cycle, cycle 3 day 22 and 29, week 14, and at 6 and 12 months were tested in this ELIspot assay format. Complete per-patient data on ELIspot

3

were presented graphically. In addition, the mean of adjusted readings across multiple time points was calculated: before vaccination (screening and C1D1); during vaccination (C1D8, C2D1, C2D8 and C3D1) and post vaccination (C3D8, C3D22 and C3D29).

ELIspot Statistical Analysis

To enable data to be summarized across all trial patients, recognition and response of individual epitopes were also defined as binary outcomes. In an unplanned analysis, vaccine immunogenicity and a relationship of size of response to vaccine dose were explored using all patients from both UK and Hong Kong trials with evaluable data from ELIspots targeting the LMP2 and EBNA1 Pepmix. Repeated measures ANOVA was used to test whether mean adjusted readings at three time-points differed significantly. For antigens in which mean adjusted readings at the three time-points were significantly different, adjusted readings at each later time-point were compared to those for pre-vaccination samples using Dunnett’s multiple comparison test. All statistical testing was undertaken using GraphPad Prism 5.

Multi-parametric flow cytometry to measure broad subsets of cells

PBMC were recovered, washed twice in cold phosphate buffered saline (PBS) and stained on ice for 30 minutes with the following monoclonal antibodies (mAbs): v450-conjugated anti-CD3 (eBioscience, San Diego, California), APC-conjugated anti-CD4 (Becton Dickinson, San Jose, California), PE-Cy7-conjugated anti-CD25 (Becton Dickinson). LIVE/DEAD fixable dead cell stain (Invitrogen, Carlsbad, California) was added to allow dead cells to be excluded from the subsequent data analysis. After staining with fluorochrome-conjugated mAbs, cells were washed twice then fixed and permeabilized using a Fixation/Permeabilization kit specifically designed for FoxP3 detection (eBioscience). After blocking with normal rat serum (eBioscience) cells were divided into two aliquots and stained with either PE-conjugated anti-FoxP3 mAb (clone PCH101, eBioscience) or the corresponding PE-conjugated isotype control mAb. Cells were washed twice with permeabilization buffer before flow cytometry. In parallel, cells were surface stained using the following mAbs: v450-conjugated anti-CD3, APC-conjugated anti-CD4, PerCP-Cy5.5-conjugated anti-CD8 (Becton Dickinson), APC-efluor780-conjugated anti-CD27 (eBioscience) and FITC-conjugated anti-CD28 (Becton Dickinson). These cells were co-stained with LIVE/DEAD fixable stain to allow dead cells to be excluded.

Cells were analyzed on an LSRII flow cytometer (Becton Dickinson). Compensation was performed using compensation beads (Invitrogen) stained with each of the above fluorochrome-conjugated mAbs; beads stained with an ECD-conjugated mAb were used as a compensation control for the LIVE/DEAD stain. Data was processed using FACSDiva software (Becton Dickinson). Forward scatter versus side scatter was used to identify PBMCs. Dead cells and CD3 negative cells were then excluded from the analysis by appropriate gating. Regulatory T cells were identified as CD4+ CD25hi FoxP3+ cells and are expressed as a percentage of CD4+ T cells.

Multi-parametric flow cytometry to measure immune response quality and phenotype of antigen-specific T-cells

Cryopreserved PBMCs were thawed and stimulated for 6hr with antigen pepmixes in the presence of 1l/ml brefeldin A (Sigma-Alrich), 0.35l/ml monensin (BD Biosciences) and, where indicated, 3l/ml CD107a-FITC (BD Pharmingen). Cells were then washed and stained

4

sequentially with Aqua viability dye (Life Technologies) and the following surface-directed monoclonal antibodies (mAbs): (i) CD14-V500 and CD19-V500 (BD Horizon); (ii) CD3-APC-H7 (BD Biosciences); (iii) CD4-Cy5.5-PE (Life Technologies); (iv) CD8-BV711 (BioLegend); and (v) CD27-Cy5-PE and CD45RO-ECD (Beckman Coulter). After fixation and permeabilization (Cytofix/Cytoperm; BD Biosciences), cells were further stained intracellularly with the following mAbs: (i) IFN-V450 (BD Horizon); and (ii) IL-2-APC, MIP-1-PE and TNF-PE-Cy7 (BD Pharmingen). Stained cells were acquired on a custom-built 20-parameter FACS AriaII flow cytometer (BD Biosciences). Data were analysed with FlowJo software v9.7.5 (TreeStar Inc.). The following sequential gating strategy was applied: (i) single cells; (ii) Boolean exclusion of fluorochrome aggregates and artefacts; (iii) live CD3+ cells; (iv) CD4+ or CD8+ cells; and (v) any of the five functional parameters or phenotypic analysis of CD27/CD45RO expression. SPICE software v5.32 was used for combinatorial analysis and data presentation (25).

5

SI Figure 1. Flow chart detailing dose escalation schema

6

SI Figure 2. MVA-EBNA1/LMP2 Vaccination does not alter broad subsets of immune cells

Healthy Donor UK0106 UK0207UK0101

CD8

CD4

0.1

3.21.1 13.3

0.4

24.5

4.9

2.42.3

0.3

1.31.4

Foxp3

CD4

A. Graphs shown change over time of immune subsets for two patients vaccinated with MVA-EL. Upper panels show lymphocyte counts, which were measured at every time point. Lower panels summarise flow cytometry analysis of PBMCs from selected timepoints before, during and after the course of vaccination. T-regs indicates the percentage of FoxP3-positive cells in the live CD3+ CD4+ lymphocyte population. The frequencies of the remaining subsets are expressed as percentages of the total live CD3+ lymphocyte population. Patient 0106 had an increased frequency of CD4+ CD8dim lymphocytes. Patients 0207 had a large increase in CD8dim and double negative cells with concomitant decreases in

7

the frequency of CD8+ and CD4+ T cells. All subsets were stable over time. Arrows underneath the x-axis indicate time of MVA-EL vaccination. B. Representative flow cytometry plots for a healthy donor and NPC patients UK0101, UK0106 and UK0207. Upper panels show percentage of CD4+ CD8dim, CD8dim and double negative cells within the live CD3+ lymphocyte population. Lower panels show the frequency of FoxP3+ regulatory T cells as a percentage of CD4 T cells.

8

SI Table 1 MHC class I and II restricted epitope peptides in EBNA1 and LMP2 used as targets in ELIspot assays

The amino acid sequence of the epitope peptides used in ELIspot assays (corresponding to the prototypic B95.8 strain of EBV) and the corresponding sequence within the MVA-EL encoded EBNA1/LMP2 fusion protein (generated from strains of EBV circulating in the Chinese population) are shown. The amino acids within brackets in the vaccine sequence are those that differ between the B95.8 and Chinese EBV sequences (B95.8 sequence Chinese sequence).

9

SI Table 2 Summary of EBNA1- and LMP2-specific immune responses stimulated by vaccination of UK NPC patients

Patient

Targets tested

Epitope targets recognised (frequency of responders / million PBMC)

HLA type Recognition above background

before vaccination

Vaccine response: ≥2-fold amplification compared to adjusted mean readings before vaccination.

HLA class I HLA class II Antigen Epitopes After cycle 2 After cycle 3

UK0101

A1, A02.01,

B7, B8

DR14, DR17,

DR52a, DR52b DQ2, DQ5

LMP2: FLY, LLW, CLG,

Pepmix CLG (380),

Pepmix (77) Pepmix (162)

EBNA1: YNL, PQCR,

Pepmix YNL (350) Pepmix (113)

UK0102

A02.01, A11,

B35, B44

DR1, DR4,

DR53, DQ3, DQ5

LMP2: FLY, LLW, CLG,

Pepmix CLG (54),

Pepmix (158) Pepmix (451)

FLY (64), CLG (137), Pepmix (753)

EBNA1: HPV, TSLY,

Pepmix Pepmix (63)

TSLY (176), Pepmix (263)

TSLY (142), Pepmix (633)

UK0103

A24, A26, B51, B52

DR14, DR15, DR52, DR51,

DQ5, DQ6

LMP2: TYG, PYL+VMS TYG (254)

EBNA1: [LRAL+MVFL],

[TSLY+PQCR], NPKF [LRAL+MVFL] (136),

[TSLY+PQCR] (364), NPKF (415) [TSLY+PQCR] (738)

UK0104

A02.01, A24,

B7, B51

DR3, DR4,

DR52a, DR53, DQ2, DQ3

LMP2: Pepmix Pepmix (854)

EBNA1: Pepmix Pepmix (132)

UK0105

A02.01, A32,

B15, B44

DR1, DR7,

DR53, DQ2, DQ5

LMP2: FLY, LLW, CLG,

Pepmix Pepmix (312) LLW (66) FLY (236), LLW (80)

EBNA1: VLK, LRAL+NPKF, TSLY, [PQCR+MVFL], VFLQ,

Pepmix Pepmix (366) VFLQ (70)

UK0106

A1, B57

DR7 DQ3

LMP2: Pepmix Pepmix (61)

EBNA1: NPKF, LRAL, PQCR, TSLY,

MVFL, Pepmix Pepmix (93)

10

UK0207

A24; A33; B13;

B58; Cw03; Cw04

DR1, DR13, DR52, DQ5,

DQ6

LMP2: Pepmix Pepmix (73)

EBNA1: Pepmix Pepmix (74) Pepmix (211)

UK0208

A02.01, A32, B40, B44

DR4, DR53, DQ3

LMP2: FLY+IED, [CLG+LLW],

Pepmix Pepmix (303)

EBNA1: VLK, LRAL+MVFL,

[TSLY+PQCR], NPKF, Pepmix

Pepmix (80)

UK0209

A02.01; A32;

B27; B44; Cw*02; Cw*07

DR3, DR16,

DR52, DR51, DQ2, DQ5

LMP2: [LLW+RRR], FLY, CLG FLY (117)

EBNA1: [TSLY+NPKF],

[PQCR+LRAL+MVFL], [VLK+LLS+LTAG]

UK0310

A24; A68; B18;

B51; Cw01; Cw05

DR3 DR13

DR52a DQ2 DQ6

LMP2: PYL, TYG, Pepmix

TYG (273), Pepmix (67)

Pepmix (350) PYL (274)

EBNA1: MVFL, TSLY, LRAL,

PQCR, NPKF, Pepmix

Pepmix (126) Pepmix (276)

UK0311

A03; B15; B44;

Cw03; Cw16

DR3, DR15,

DR51, DR52, DQ2, DQ6

LMP2: WTL

EBNA1: MVFL+VFLQ, [NPKF+TSLY], [LRAL+PQCR]

UK0212

A1, A32, B44, B51,

DR4, DR13,

DR52, DR53, DQ3, DQ6

LMP2: Pepmix Pepmix (122) NOT DONE NOT DONE

EBNA1: PQCR, LRAL

Pepmix Pepmix (392) NOT DONE NOT DONE

UK0313

A02.07; B46;

Cw*01;

DR9, DR12,

DR52, DR53, LMP2:

FLY, LLW, CLG, [LLS+LTAG]

11

DQ3 EBNA1:

VLK, [TSLY+NPKF], [LRAL+PQCR+MVFL]

[TSLY+NPKF] (107)

UK0414

A02.01; A33;

B40; B58; Cw*03; Cw*07

DR3, DR16,

DR51, DR52b, DQ2, DQ5

LMP2: FLY, LLW, CLG, IED,

[LLS+LTAG], Pepmix

FLY (325), LLW (357), CLG (266), IED (339)

EBNA1: [TSLY+NPKF],

[LRAL+PQCR+MVFL], Pepmix

Pepmix (91)

UK0415

A01 B08

Cw07

DR3 DR52a DQ2

LMP2

EBNA1 YNL, VFLQ, [TSLY+NPKF],

[PQCR+LRAL],

Cryo-preserved PBMC from multiple time points were thawed and tested in ELIspot assays at 3x105/well against single peptides or groups of peptides (marked in square brackets) or against the DMSO solvent control. MHC class I and II epitopes are denoted by 3 and 4 residue codes respectively. Also shown are pepmixes, 15mer peptides spanning the entire primary sequence of EBNA1 or LMP2. Adjusted readings were the difference between the spot forming cells / well of the test and control wells. The mean of adjusted readings across multiple time points was calculated: before vaccination (screening and cycle 1 day 1 - C1D1); after cycle 1 (C1D8, C2D1, C2D8 and C3D1) after cycle 2 (C2D8 and C3D1) and after cycle 3 (C3D8, C3D22 and C3D29). Antigen recognition is defined as an adjusted reading (difference between test and control wells) exceeding 10sfc/300,000 PBMC and ratio of test versus control wells ≥2.0. Amplification of response during or post vaccination or during follow up is defined as antigen recognition in which the ratio compared to the pre-vaccination adjusted reading is ≥2.0. Frequencies, converted in this table to responders/million PBMC, are shown only for antigen recognition detected before vaccination or for amplified responses during or post vaccination or during follow up.

12

SI Figure 3. EBNA1 antibody response before and after MVA-EL vaccination.

The EBNA1 IgG antibody level before, during and after the course of vaccination is shown for each

patient. One patient was vaccinia seronegative before vaccination and is represented by an open symbol at each timepoint. Using data for all patients, levels of prevaccination anti-EBNA1 IgG titre were compared to those after one, two or three vaccinations. The mean values (horizontal bars) were significantly different (P=0.025 repeated measures ANOVA). Compared with the prevaccination control sample, the mean difference in reading was significant at the 5% level only for the samples taken at C3D8.