A phase I vaccination study with tyrosinase in patients with stage II melanoma using recombinant modified vaccinia virus Ankara (MVA-hTyr)

Abstract

A significant percentage of patients with stage II melanomas suffer a relapse after surgery and therefore need the development of adjuvant therapies. In the study reported here, safety and immunological response were analyzed after vaccination in an adjuvant setting with recombinant modified vaccinia virus Ankara carrying the cDNA for human tyrosinase (MVA-hTyr). A total of 20 patients were included and vaccinated three times at 4-week intervals with 5×10⁸ IU of MVA-hTyr each time. The responses to the viral vector, to known HLA class I–restricted tyrosinase peptides, and to dendritic cells transfected with tyrosinase mRNA, were investigated by ELISpot assay on both ex vivo T cells and on T cells stimulated in vitro prior to testing. The delivery of MVA-hTyr was safe and did not cause any side effects above grade 2. A strong response to the viral vector was achieved, indicated by an increase in the frequency of MVA-specific CD4⁺ and CD8⁺ T cells and an increase in virus-specific antibody titers. However, no tyrosinase-specific T-cell or antibody response was observed with MVA-hTyr in any of the vaccinated patients. Although MVA-hTyr provides a safe and effective antigen-delivery system, it does not elicit a measurable immune response to its transgene product in patients with stage II melanoma after repeated combined intradermal and subcutaneous vaccination. We presume that modification of the antigen and/or prime-boost vaccination applying different approaches to antigen delivery may be required to induce an effective tyrosinase-specific immune response.

Introduction

The incidence of malignant melanoma is rapidly increasing. Melanoma now accounts for approximately 4.1% of all cancers in the United States [23]. The death rate due to melanoma has doubled during the last 35 years. At present, surgical resection is the major curative therapy for local melanoma (American Joint Committee on Cancer [AJCC] stages I and II). Metastatic disease (AJCC stages III and IV) is fatal in most cases, with a small proportion of patients surviving long-term after chemoimmunotherapy [30]. Only patients with stage I disease have a good chance to be cured, whereas in stage II melanoma (>1.0 mm), relapse rates range from 10% to 40% with a mortality rate of 30–40% within 5 years.

Melanoma is considered highly responsive to the attacks of the immune system. For instance, vitiligo, either appearing spontaneously or in the course of immunotherapy, was associated with a survival advantage [14, 42]. A rather small, albeit consistent, proportion of patients with metastatic disease respond to the T-cell growth factor IL-2 or to vaccination with peptides [34, 41, 44]. Along the same line of evidence, antimelanoma T cells have been isolated from lesions of regressing melanomas [61]. Different categories of melanoma-associated antigens (MAAs) have been described that can be recognized by autologous T lymphocytes (for review, see [43]). Among these, tyrosinase represents a prototypic antigen for melanosomal differentiation antigens (MDA), and is expressed in melanomas as well as in normal melanocytes. It is the key enzyme of melanin synthesis and has been well characterized in its function as target for tumor-reactive lymphocytes [3, 44, 59]. Various immunogenic peptides of tyrosinase presented by either class I or class II HLA molecules have been identified, making tyrosinase an ideal target for vaccination strategies against melanoma.

The development of DNA-based vaccination strategies using the complete tyrosinase gene expressed in patients‘ antigen-presenting cells (APCs) utilizes all known peptides and also potentially unknown peptides of tyrosinase. As recombinant vaccinia viruses are known to be very potent vectors for gene delivery, modified vaccinia virus Ankara (MVA), a virus-strain derived from vaccinia virus strain Ankara by serial passages in primary chicken fibroblast (CEF) cells, was chosen for vaccination. Although MVA does not replicate in human cell lines, transgenes are efficiently transcribed [50]. In extensive studies in Germany, MVA has been used in more than 120,000 human subjects in a two-step vaccination protocol against smallpox, combined with conventional vaccinia virus [35]. The vaccination caused only minor side effects and is therefore considered safe. In recent years, MVA has been used as a vector for the delivery of various antigens in animals [46] and humans [36]. MVA-hTyr carries a full-length human tyrosinase cDNA under the control of the MVA early/late promoter P7.5. Cells infected with this recombinant virus are recognized by human tyrosinase-specific CTL clones [58]. These findings indicate that the inserted cDNA is properly expressed and that antigenic peptides are generated.

Efficient tyrosinase- and melanoma-specific T-cell responses were induced in vitro by MVA-hTyr-infected dendritic cells (DCs) as stimulators of human peripheral blood mononuclear cells (PBMCs) [13]. When HLA-A2.1–transgenic mice were vaccinated with MVA-hTyr, they developed a strong cellular antityrosinase response, outlining the potency of the vaccine to induce such responses [13].

We conducted a phase I trial in an adjuvant setting for patients with stage II melanoma. MVA-hTyr was administered three times at 4-week intervals, and blood samples for measuring immunological response were taken immediately before and 2 weeks after each vaccination. Primary objectives of the study were safety, feasibility of the study medication, and the induction of a tyrosinase-specific immune response. Secondary objective was the induction of a MVA-specific immune response.

Material and methods

Study design and patient recruitment

The study was performed according to good clinical practice (GCP) standards, the Declaration of Helsinki, local legal regulatory requirements, and the German “Kommission für somatische Gentherapie” of the “Bundesaerztekammer”. Patients included were older than 18 years and of either gender with a good performance status (Karnofsky >80%). Stage II melanoma (based on the AJCC staging system of 1992) needed to be confirmed by histology and clinical staging. The availability of tumor tissue for histological analysis was requested, but not obligatory. Patients were positive for at least one of the following HLA alleles: HLA-A1, HLA-A2, HLA-A24, HLA-B35, or HLA-B44. Surgery for any reason was allowed, if a minimum of 4 weeks had elapsed before study entry and the patient’s recovery from all treatment-related toxicity was assured. The patients gave written informed consent and had to be able to adhere to the treatment schedule. Pregnancy was excluded in female patients prior to each application of the study medication.

Patients were vaccinated three times at 4-week intervals with MVA-hTyr at a dose of 5×10⁸ infectious units (IU) included in a total volume of 1 ml (2×0.1 ml intradermally and 2×0.4 ml subcutaneously). A volume of 80 ml of heparinized peripheral blood and 10 ml of serum was drawn prior to, and 2 weeks after, each vaccination. Study objectives were the assessment of biological safety, the induction of virus- and tyrosinase-specific T-cell and antibody responses, and clinical screening for metastases or recurrence of melanoma within the study period of 70 days. Long-term monitoring for relapse is still ongoing.

At study entry, patients were screened for metastases by means of sonography of local lymph nodes and CT scans of chest and abdomen. Seven patients had additional magnet resonance imaging (MRI) of the brain. All patients were screened for antibodies against cytomegalovirus (CMV), Epstein–Barr virus (EBV), and influenza A. A total of 12 patients had a proven history of smallpox vaccination, 1 patient had not been vaccinated, and in 7 patients, vaccination status remained unclear (Table 1).

Table 1.

Summary of patient characteristics. All patients had stage II melanoma according to the AJCC classification valid at study entry [20]. ND not documented

UPN	Gender	Age at study entry (years)	HLA class I	Prior vaccination against smallpox	Localization of primary lesions	Days from surgery to study entry	Classification
1	M	70	A2,11; B7,56; Cw1,w7	Yes	Lumbar, left	40	Desmoplastic
2	M	57	A1,25; B8, 55; Cw3,w7	Yes	Left shoulder	366	ND
					Lower right leg	418
					Occiput	366
3	M	43	A24,30; B13,18; Cw5,w6	Yes	Popliteal, right	36	ND
4	M	58	A2,3; B35,51; Cw14,w4	Yes	Supra auricular, left	29	Nodular
5	F	37	A24; B7,62; Cw3, w7	Yes	Thigh, left	38	Nodular
6	M	53	A1; B7,18; Cw7	Unclear	Left cheek	47	ND
7	F	57	A 1,2; B8,57; Cw6,w7	Yes	Gluteal region	33	ND
8	M	75	A2,33; B14,57; Cw6,w8	Yes	Thorax, right	37	ND
9	M	56	A24,29; B44,51; Cw4	Yes	Abdomen, right	110	ND
11	M	54	A2,23; B18,49; Cw5,7	Yes	Lumbar, left	119	SSM
12	M	65	A1,31; B7,14	Unclear	Back, left	335	Nodular
13	M	68	A2; B52,60; Cw3	Unclear	Interscapular	115	SSM
14	F	53	A1,3; B18,52	Unclear	Upper trunk, right	263	SSM
15	M	68	A1,24; B8,61; Cw2	Yes	Left shoulder	80	Unclassified, regressive
16	M	38	A2,32; B44; Cw5	Yes	Left lower arm	89	Nodular
21	F	58	A2; B65,B39	Yes	Right upper arm	54	SSM
22	M	29	A3,24; B7,39	No	Right lower arm	148	SSM
41	F	30	A1,2; B13,57; Cw6	Unclear	Pretibial	117	SSM
42	M	60	A2,11; B35,81	Unclear	Right foot	127	ND
43	F	46	A 1,31; B39,60; Cw3	Unclear	Left cheek	125	Nodular

Immunohistochemistry for tyrosinase in primary melanomas

Immunohistochemistry was performed on 10 μm-paraffin sections. Briefly, sections were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval using a microwave oven with 0.1 M EDTA buffer (pH 6) for 8 min. Sections were incubated with the primary antibodies (1:25 working dilution, NCL-Tyros; Novocastra Laboratories, Newcastle-upon-Tyne, UK) for 45 min at room temperature. Tyrosinase was localized via the standard avidin–biotin immunoperoxidase method with diaminobenzidine chromogen. Sections were counterstained with hematoxylin.

For semiquantitative determination of tyrosinase expression, we used a four-tier score. Lack of expression was classified as 0, weak expression as +, moderate or irregular expression as ++, and strong, uniform expression as +++. All sections were imported into Photoshop (version 6.0; Adobe, Unterschleissheim, Germany) using a standard diagnostic microscope, a video camera, and a Macintosh computer with built-in capture board (G3; Apple Computer, Cupertino, CA, USA). Determination of mean immunoperoxidase signal strength was performed in Photoshop using the magic wand tool and the histogram command as previously described in detail [32]. The mean value of a gray scale, with 0 for black and 255 for white, was thus calculated and used as a measure of tyrosinase expression.

Flow cytometric analysis

The PE-labeled HLA-A2/peptide multimer complexes were synthesised as previously described [6]. The naturally occurring Tyr_369–377 peptide [59] was used for generating HLA-A2/peptide multimers to detect Tyr_369–377-specific T cells in the peripheral blood of HLA-A2–positive patients vaccinated with MVA-hTyr. HLA-A2/peptide multimers consisting of the HLA-A2–binding HER-2_369–377 peptide [17] were chosen as negative control. Multimers were tested for specific labeling using T-cell clones BS-T5 (anti-Tyr_369–377) and SuHi-E7 (anti-HER-2_369–377) as positive controls.

A total of 4×10⁶ PBMCs were incubated with PE-labeled HLA-A2/peptide multimers for 45 min on ice in 100 μl buffer (PBS 1% BSA, pH 7.4). Then anti-CD8-FITC (Caltag Laboratories, Burlingame, CA, USA) was added. After 15 min, the cells were washed three times and resuspended in buffer with PI (2μg/ml). PI-negative cells were gated, and 500,000 events were counted by using Coulter Epics XL (Beckmann-Coulter, Brea, CA, USA).

IFN-γ ELISpot assay

Enzyme-linked immunosorbent spot (ELISpot) analysis was performed as previously described [58]. Briefly, Multiscreen HA plates (Millipore, Bedford, MA, USA) were coated with 10 μg/ml of monoclonal antihuman IFN-γ antibody (1-D1 K; Mabtech, Stockholm, Sweden). Effector cells were seeded in triplicates together with antigen-loaded antigen-presenting cells. For ex vivo analysis, PBMCs were thawed, CD4⁺ and CD8⁺ cells were isolated using immunomagnetic beads according to the manufacturer’s instructions (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany) and seeded with 1×10⁵  cells/well. Either HLA class I–transfected K562 cells (1×10⁵/well) [4] or autologous DCs (2×10⁴/well) were used as APCs. Loading with antigen was performed by either adding peptides, infecting with virus, or transfecting with mRNA.

The cells were incubated at 37°C in 5% CO₂ in a final volume of 100 μl/well X-Vivo 15 (BioWhittaker, Apen, Germany) for 20 h. Captured cytokine was labeled after incubation for 2 h at 37°C with biotinylated mAb anti-hIFN-γ (7-B6-1; Mabtech) at 2 μg/ml in PBS/0.5% BSA using an avidin-peroxidase complex (1/100; Vectastain Elite Kit; Vector, Burlingame, CA, USA). Peroxidase staining was performed with 3-amino-9-ethyl-carbazole (Sigma-Aldrich, Muenchen, Germany) for 4 min and stopped by rinsing the plates under running tap water. Spot numbers were automatically determined with the use of computer-assisted video image analysis (CVIA). Hardware and software (KS ELISpot version 4.4) of the imaging system used in this study were developed by Zeiss-Kontron (Jena, Germany).

Peptides

Peptides (see Table 2) were obtained from Affina Immuntechnik (Berlin, Germany) at a purity of >80% (immunology grade) and stored in PBS/DMSO 5% in aliquots at −20°C until use. After thawing, peptides were stored at 4°C for up to 2 weeks.

Table 2.

Peptides used for T-cell response analysis

Protein	Peptide	Residues	HLA restriction	References
Tyrosinase	SSDYVIPIGTY	146–156	HLA-A1	[26]
Tyrosinase	KADICTDEY	243–251	HLA-A1	[28]
Tyrosinase	MLLAVLYCL	1–9	HLA-A2	[59]
Tyrosinase	YMDGTMSQV	369–377	HLA-A2	[48, 59]
Tyrosinase	AFLPWHRLFL	206–215	HLA-A24	[25]
Influenza BP	VSDGGPNLY	591–599	HLA-A1	[12]
CMV-pp65	NLVPMVATV	495–503	HLA-A2	[11]
EBV-BMLF1	GLCTLVAML	259–267	HLA-A2	[7]
HIV-RT	ILKEPVHGV	476–484	HLA-A2	[55]
HER-2/neu	KIFGSLAFL	369–377	HLA-A2	[17]

Generation of dendritic cells

The generation of immature DCs was performed as described [24]. Briefly, peripheral blood monocytes were isolated from PBMCs by adherence to plastic at 1.5×10⁷  cells/well in a six-well plate in 3 ml X-Vivo 15 for 45 min on day 0. Nonadherent cells, containing large amounts of lymphocytes, were removed by repeatedly washing with PBS. X-Vivo 15 (3 ml) was added to the adherent cells. On day 1, the medium was replaced by 3 ml X-Vivo 15 containing 800 IU/ml GM-CSF (Leucomax, Novartis, Nürnberg, Germany) and 1,000 IU/ml IL-4 (Strathmann Biotech, Hannover, Germany). On day 3 and on day 5, 1 ml of this medium was replaced by X-Vivo 15 containing 1,600 IU GM-CSF/ml and 1,000 IU IL-4/ml. After 6 days of culture, immature dendritic cells were harvested by repeatedly washing with cold PBS. In previous experiments, we had confirmed that the use of immature DCs in this setting was equivalent to the use of mature DCs.

Infection of dendritic cells with viruses

The study medication (MVA-hTyr) as well as MVA-hTyr and wild-type MVA for in vitro analyses was obtained from Bavarian Nordic (Martinsried, Germany) and stored at a concentration of 5×10⁸ IU/ml. Immature DCs were seeded at 0.7–1×10⁶  cells/well in a 24-well plate and incubated overnight in X-Vivo 15 containing IL-4 and GM-CSF. Infection was performed in 200 μl of X-Vivo 15 containing MVA at a multiplicity of infection (MOI) of 10. After 1 h, 1.8 ml X-Vivo 15 was added to each well. Infected DCs were harvested after an additional incubation period of 2 h.

Optimal recognition of virus-infected DCs was verified using the Tyr_369–377-specific T cell line IVSB [59]. Best responses were achieved, when MVA-infected immature DCs were used immediately after infection (data not shown).

In vitro stimulation with peptides

To expand peptide-specific T cells in vitro, PBMCs were stimulated as described [9] with minor modifications. Briefly, PBMCs were loaded with 20 μg/ml peptide in AIM-V medium (Invitrogen, Karlsruhe, Germany) containing 1% HS for 1 h at 37°C, washed, and resuspended in AIM-V containing 5% HS, 10 ng/ml IL-7 (R&D Systems), 20 IU/ml IL-2 (Roche), and 10 ng/ml IL-4. The percentages of CD8⁺ cells at the start of the culture were calculated using flow cytometry. After 7 days, fresh culture medium was added containing 20 μg/ml of peptide. Cultures were split if necessary.

“Bulk” and limiting dilution (LD)-like cultures

Stimulations starting with 3×10⁶ unseparated PBMCs (“bulk” cultures) were used to evaluate the existence of any expandable CD8⁺-precursor T cells. To estimate precursor frequencies of peptide-specific CD8⁺ T cells, limiting dilution(LD)-like cultures were conducted.

LD-like cultures [9] were started in two concentrations in 96-well round-bottom plates. On day 14, the content of each well was split into four wells of an IFN-γ ELISpot assay with peptide-loaded K562 cells, which had been transfected with the appropriate HLA class I allele. Cultures were considered positive, when the mean peptide-specific spot number was significantly above background response to unloaded HLA-transfected K562 cells.

Plasmids

Full-length tyrosinase cDNA was cloned in pGEM3z (Promega, Mannheim, Germany) under the control of a T7 promoter (designated pGEM3z-hTyr), transformed into E.coli Top 10F and purified using the Qiagen Endofree Plasmid Maxi Kit (Qiagen, Hilden, Germany). The expression vector pcDNA6/V5-HisB-pp65 (BamHI-BamHI) was kindly provided by Dr B. Plachter (Institute of Virology, Universität Mainz, Germany).

Transfection of dendritic cells with in vitro–transcribed mRNA

The plasmid pGEM3z-hTyr was linearized with Hind III (Roche) and pcDNA6/V5-HisB-pp65 was linearized with Xho I (Roche). For in vitro transcription the Ribomax large-scale RNA production system T7 (Promega) was used according to the manufacturer’s instructions. Briefly, linearized DNA templates were incubated for 4 h at 37°C in a reaction mix containing T7 5x reaction buffer, ATP, CTP, UTP (7.5 mM each), 1 mM GTP, 5 mM m7 G(5’)ppp(5’)G cap-analogue (Invitrogen), and T7 RNA polymerase. The in vitro–transcribed RNA was further polyadenylated using poly A polymerase (USB, Cleveland, OH, USA) according to the manufacturer’s instructions. Briefly, the RNA was incubated for 1 h at 30°C in a reaction containing poly A polymerase reaction buffer, ATP (Promega; 600 pmol/pmol of RNA) and poly A polymerase (2.5 U/pmol RNA).

Dendritic cells were transfected as described [56] with minor modifications [5]. Briefly, immature DCs were harvested, washed twice, and adjusted to final cell densities of 2.5–5×10⁶ cells/ml in Optimem (Gibco BRL, Life Technologies, Karlsruhe, Germany). Two hundred microliters of the cell suspension was preincubated in a 0.4-cm-gap electroporation cuvette on ice, 45 μg of clonal mRNA was added, and cells were immediately pulsed at 350 V and 300 μF (Gene Pulser II, BioRad, München, Germany). Afterwards, cells were reseeded in culture medium, counted, and added to the ELISpot assay at 2×10⁴ DCs per well. Antigen expression of transfected DCs was verified using the Tyr_369–377–specific T cell line IVSB [59] or CD8-positive T cells of CMV-positive healthy donors with a documented response to CMV-pp65_495–503.

Detection of antibody responses to MVA

Modified vaccinia virus Ankara–specific IgG antibody responses were measured using a direct enzyme-linked immunosorbent assay (ELISA). The 96-well Transp Maxis plates (Nunc, Wiesbaden, Germany) were coated overnight at 4°C with 100 μl (7.5 μg/ml) of MVA-infected CEF cell lysate in coating buffer (200 mM Na₂CO₃, pH 9.6). As control, plates were simultaneously coated with 100 μl (7.5 μg/ml) of uninfected CEF cell lysate in coating buffer (200 mM Na₂CO₃, pH 9.6). After blocking with PBS–FCS 5% (PAA Laboratories, Linz, Austria), test sera were titrated in duplicates using twofold serial dilutions and incubated for 1 h at room temperature. A goat antihuman horseradish peroxidase (HRP) antibody (1:40,000; Sigma-Aldrich) was used for detection. Staining was performed using o-phenylenediamine dihydrochloride tablets (Sigma-Aldrich) prepared according to the manufacturer’s instructions, and the reaction was stopped by adding 50 μl of 3 M H₂SO₄ (Merck, Darmstadt, Germany) per well. The optical density (OD value) was read at 492 nm, with a reference of 405 nm. The antibody titers were calculated by linear regression and defined as the serum dilution that resulted in an OD of 0.3.

Detection of antibody responses to tyrosinase

Recombinant tyrosinase fragment was generated as described previously [52]. Briefly, the cDNA coding for a 28-kDa domain of human tyrosinase (aa 285–475) was cloned into expression vector pQE30 (Qiagen) enabling expression with N-terminal 6×His-tag. The recombinant tyrosinase fragment was overexpressed in E.coli M15 [pREP4] (Qiagen), purified from cell lysates by affinity chromatography on a Ni-NTA–agarose column and refolded by dialysis against buffer RFB (100 mM H₃BO₃, 10 mM Tris, pH 7.4). The purified recombinant His-tagged fragment was diluted in 50 mM bicarbonate buffer (pH 9.5) to a final protein concentration of 5 μg/ml as determined by the Bradford assay (BioRad). The tyrosinase solution was dispensed at a concentration of 5 μg/ml into 96-well plates (100 μl/well) and incubated overnight at 4°C. After removal of the protein solution, plates were blocked with 5% skim dry milk solution (Sigma-Aldrich) in PBS for 1 h at 37°C (100 μl/well). Serum samples diluted 1:100 in PBS were added at 100 μl per pre-coated well. The monoclonal antityrosinase antibody 9A1 [52] was used as positive control. Detection was performed with 1:100,000 diluted goat antihuman IgG F(ab’)2 labeled with HRP (Coulter Immuno-Diagnostics, Krefeld, Germany) and developed with tetramethylbenzidine (TMB) solution (1 mg TMB; Sigma-Aldrich) diluted in 10 ml 0.05 M citrate buffer (pH 5.0) with 0.006% H₂O₂. The reaction was stopped with 100 μl of 0.25 M H₂SO₄, and the absorbance at 460 nm was measured. All samples were run in duplicates on the plates. Sera from the 20 melanoma patients and from 89 healthy donors were tested simultaneously.

Results

Patient recruitment

A total of 20 patients from four German centers (Erlangen, Essen, Mainz, and Munich) were included in the study. Of these patients, 6 were female and 14 were male. The age at study entry ranged from 29 to 75 (53.7 ± 13 years; mean ± SD), and the average time span between surgery and study entry was 102.2 days (±80.6; range 33–418). One patient (patient 2) was excluded from the study after the second vaccination due to local recurrence of the melanoma at the excision site. Table 1 summarizes the data of the patients included.

Safety and reactogenicity

A total number of 316 adverse events were reported, of which 278 were classified as being at least possibly related to application of the study medication. Of these, 77% were application site problems. All patients experienced local rash and injection site pain. Some patients developed an inflammatory reaction and indurations of the skin, and even, less frequently, some patients reported local pruritus, bruising, and edema. All local adverse reactions were classified as mild or moderate. General symptoms were commonly mild. Most frequently fatigue (six patients), fever (five patients), headache (four patients), and “influenza-like symptoms” (four patients) with myalgia occurred. One patient had fever and hypertension after the first vaccination. For safety reasons, this patient was hospitalized during the second vaccination, but showed no further systemic reactions. Other events such as anorexia, diarrhoea, flatulence, nausea, tooth ache, myalgia, back pain, migraine, cough, dyspnoea, lymphadenopathy, and herpes simplex virus I reactivation were reported by isolated patients following vaccination. The local relapse of melanoma in patient 2 was the only serious adverse event, but was assessed as unlikely to be related to the vaccination. In conclusion, the investigational drug was safe and well-tolerated; local reactions and general symptoms were mild and occurred only temporarily.

Tyrosinase expression of primary tumors

Tyrosinase expression in primary tumors was analyzed by immunohistochemistry. The primary tumors of 15 patients were available for this analysis. The level of tyrosinase expression was estimated by Photoshop-based analysis of chromogen distribution [31] and semiquantitatively classified as negative, weak (+), moderate (++), or strong (+++).

All tumors examined expressed tyrosinase. Only the tumor of patient 14 had a strong expression (gray scale 89), the tumors of seven patients showed a moderate (mean gray scale 117.9), and the tumors of seven patients showed a weak patchy expression of tyrosinase (mean gray scale 148.1). The primary tumor of patient 2 was found to have the weakest expression of tyrosinase. This patient experienced a local recurrence of the melanoma and exhibited a stronger expression of tyrosinase in the relapsed tumor than in the primary lesion (gray scale value 126 vs 159).

Humoral response to MVA and tyrosinase

The humoral response to MVA was measured with direct ELISA in the sera of all 20 patients. Of the 12 patients with a proven history of vaccination against smallpox (patients 1–5, 7–9, 11, 15, 16, and 21), 6 had no preexisting anti-MVA antibody response (patients 1, 3, 4, 7, 9, and 16). Patient 22 had not been vaccinated against smallpox before and, accordingly, he had no preexisting anti-MVA antibody response. Of the remaining seven patients, whose vaccination status was unclear, only patient 14 had no preexisting antibody titer to MVA. All patients exhibited a strong increase of anti-MVA antibody titers after vaccination with MVA-hTyr. Those patients with preexisting titers showed a 2.8-fold (patient 15) to 36.4-fold (patient 41) increase. The data of the humoral response to MVA are summarized in Fig. 1. The humoral response to tyrosinase was investigated by introducing the hydrophilic part of the tyrosinase molecule (aa 285–475) as antigen in a direct ELISA. The sera of all patients and the sera of 89 healthy individuals were analyzed simultaneously. No antibodies above background were detected before or after vaccination in any of the patients (data not shown).

Fig. 1.

MVA-specific antibody response. Antibody titers toward MVA were determined by direct ELISA. Titers were calculated by linear regression and defined as the serum dilution that resulted in an OD of 0.3. Eight patients were anti-MVA sero-negative at study entry, which is indicated by a titer of one. Days of blood samples are displayed. Gray bars indicate mean titers of all patients.

Ex vivo CD8⁺ and CD4⁺ T-cell responses to MVA and MVA-hTyr

T-cell responses to MVA were measured in parallel on ex vivo CD8⁺ and CD4⁺ lymphocytes with IFN-γ ELISpot assays. We used infected autologous immature DCs as stimulators. Nine patients were analyzed for anti-MVA CD8⁺ T-cell responses. Seven of them (patients 1–5, 8, and 21) had a documented history of smallpox vaccination during their childhood. The two others, patients 13 and 43, had probably been vaccinated before, according to their positive anti-MVA antibody titers. All patients showed an increase in the frequency of MVA-specific HLA class I–restricted T cells after vaccination (Fig. 2a). Individual maximum frequencies ranged from 0.15% in patient 43 to 1% in patient 1. Only patients 3 and 5 had a clear anti-MVA CD8⁺ response already on day 0. In these two patients, frequencies increased by 14.5-fold and 13-fold, respectively, during the vaccination period. For patients 2, 3, and 4, the frequencies of MVA-specific CD8⁺ T cells were determined in all available samples. Regularly, the frequencies of anti-MVA CD8⁺ cells rose approximately tenfold after the first vaccination (day 14 vs day 0) and dropped approximately by half until day 28. The frequencies evaluated 2 weeks after the second vaccination (day 42) exceeded those from day 14 but remained stable throughout the remaining observation period (data not shown).

Fig. 2.

MVA- and MVA-hTyr–specific T-cell responses. The frequencies of CD8⁺ (a) and CD4⁺ (b) T cells (100,000/well) in response to MVA-hTyr-infected (black columns) and to MVA-infected (gray columns) autologous DCs (20,000/well) were determined in a 20-h IFN-γ ELISpot assay. Mock-infected DCs (white columns) served as negative controls. Days of blood sampling are indicated. Data represent means and standard deviations of triplicates; # indicates unique patient number.

In none of the patients did the frequencies of CD8⁺ T cells against MVA-hTyr clearly exceed those measured against unmodified MVA. In patients 1, 2, and 4, the frequencies against MVA-hTyr were even lower. The latter finding, however, was caused by a loss in viral titer in a single aliquot of MVA-hTyr.

Seven patients were tested for CD4⁺ T-cell responses to MVA and MVA-hTyr. Five patients (patients 1, 4, 5, 13, and 21) had a clear, preexisting CD4⁺ response to MVA. In all seven patients, an increase in frequency of spot-producing cells to MVA was found. Again, the frequencies of T cells against MVA-hTyr did not exceed those against MVA. Patient 43 developed a CD4⁺ T-cell response, which was just at the detection level. This patient also had a low CD8⁺ T-cell response. Except for patient 4, CD4⁺ T-cell responses were lower than those of CD8⁺ T cells against MVA (Fig. 2b).

Ex vivo CD8⁺ T-cell response to known peptides of human tyrosinase

Out of 20 patients, 10 were HLA-A2–positive. Ex vivo CD8⁺ T cells of these patients were tested with peptide-loaded K562/A*02011 cells, except for patient 4, whose CD8⁺ T cells were tested with peptide-loaded autologous DCs. For those patients who were serologically EBV-positive, the HLA-A2–restricted EBV peptide BMLF-1_259–267 was used as a positive control (Fig. 3). No response to the two known HLA-A2–restricted peptides hTyr_1–9 and hTyr_369–377 was observed in any of the patients before or after vaccination with MVA-hTyr.

Fig. 3.

T-cell response to tyrosinase peptides. The frequencies of CD8⁺ T cells responsive to K562/A2 cells (75,000/ well) or autologous DCs (patient 4; 20,000 DC/well) loaded with hTyr_1-9 (solid circles) or hTyr_369-3377 (solid squares) were determined in a 20-h IFN-γ ELISpot assay. Unloaded APCs (solid triangles) served as negative controls. APCs loaded with BMLF1_259–267 (open circles) served as positive controls in patients seropositive for EBV. Please note that a different y-scale was used for data of patient 11. Data represent means and standard deviations of triplicates. # indicates unique patient number.

Eight out of nine HLA-A1–positive patients were tested with the two HLA-A1–restricted tyrosinase peptides hTyr_146–156 and hTyr_243–251 (not shown). The influenza A basic protein peptide PB_591–599 was chosen as positive control. Ex vivo CD8⁺ T cells from PBMCs of patients 4, 12, 14, 15, and 43 were tested with K562/A*0101 cells; ex vivo T cells from patients 2, 6, and 7 were tested with autologous DCs as APCs. No significant reactivity to HLA-A1–restricted tyrosinase peptides was detected.

The response to the only known HLA-A24–restricted tyrosinase peptide (hTyr_206–215) was tested in all five HLA-A24–positive patients. Samples were tested either with autologous dendritic cells (patient 3) or with K562/A*2402 cells as APCs. Again, no HLA-A24–restricted antityrosinase response was found (not shown).

HLA-A2 peptide multimer staining

Patients 11, 13, and 16 were analyzed using HLA-A2 peptide multimers containing the human tyrosinase peptide hTyr_369–377 at three different time points each. The HER-2_369–377 peptide served as a negative control in these experiments. T-cell clones generated from healthy donors recognizing the respective peptide, served as a positive control. In none of these three patients’ PBMCs was specific tetramer staining detectable (data not shown).

Response to DCs transfected with clonal tyrosinase mRNA

To avoid background reactivity to viral vectors, clonal mRNA was chosen as an antigen format to screen for specific T cells. Full-length hTyr mRNA and, as a control, CMV pp65 mRNA were transcribed in vitro from plasmids. After in vitro transcription and capping, RNA was polyadenylated and electroporated into autologous DCs. A total of 12 patients were analyzed with mRNA-electroporated DCs. None exhibited a tyrosinase-specific response of CD8⁺ or CD4⁺ T cells (Fig. 4), whereas clear CD4⁺ and CD8⁺ responses to CMV pp65 were found in patients 4, 8, 14, and 21. CD4⁺ T-cell reactivity was only found, if a CD8⁺ T-cell response also existed. Again, the frequencies of CD8⁺ pp65-specific T cells exceeded those of CD4⁺ T cells. T-cell reactivity to pp65 correlated well with the CMV serostatus, except for in the case of patient 22.

Fig. 4.

T-cell response to DCs transfected with tyrosinase and CMV-pp65 mRNA. The frequencies of CD8⁺ (a) and CD4⁺ (b) T cells (100,000/well) in response to autologous dendritic cells (20,000/well) transfected with mRNA encoding human tyrosinase (black columns) or CMV-pp65 (gray columns) was determined in an IFN-γ ELISpot assay. Mock-transfected DCs (white columns) served as negative controls. Please note that the y-scale is different for the data of patients 4, 8, and 14 (left lane). The data of a 24-h ELISpot assay represent mean and standard deviations of triplicates. # indicates unique patient number.

In vitro expansion of peptide-specific CD8⁺ lymphocytes

In an attempt to uncover low frequencies of specific CD8⁺ T cells that were not detectable by ex vivo tests, PBMCs were stimulated with HLA class I–restricted hTyr peptides in vitro. Stimulation was performed in either bulk cultures or in LD-like assays (see “Methods”). PBMCs of nine HLA-A2–positive patients were stimulated with hTyr_369–377 in LD-like assays, and PBMCs of four HLA-A24–positive patients were stimulated with hTyr_209–217 in bulk cultures. After two rounds of stimulation, no tyrosinase-specific T cells were detectable. PBMCs of seven HLA-A1–positive patients were stimulated with hTyr_146–156 and hTyr_243–251 either in bulk cultures (patients 2, 14, 15, and 43) or in LD-like assays (patients 6, 12, and 41). Patient 6 exhibited positive cultures for both HLA-A1–restricted peptides. Frequencies ranged from 3.8×10⁻⁶ to 3×10⁻⁵ for hTyr_243–251 and 9.6 ×10⁻⁶ to 9.35×10⁻⁵ for hTyr_146–156. This response to tyrosinase was already present before and did not increase after vaccination (not shown).

Discussion

Tyrosinase, like other melanosomal differentiation antigens, is targeted by the humoral and the cellular immune system. This has been demonstrated in melanoma patients as well as in healthy individuals [22, 25, 26, 28, 33, 59]. Induction of a tyrosinase-specific immune response by therapeutic measures leads to tumor protection in animal models [37] and has been associated with tumor regression in some melanoma patients [1, 29]. Therefore, tyrosinase has become a major target for vaccination efforts in melanoma patients. Recent studies have focused mainly on vaccination with peptides that are recognized by CD8⁺ T cells and have shown encouraging results [29, 44]. The use of peptides is limited to known epitopes and to the presence of their restricting HLA alleles. This limitation can be overcome by a genetic vaccine that leads to the expression of full-length tyrosinase in patients’ APCs and to the presentation of all peptides that can be presented by the individual HLA molecules of the patients. This procedure takes best advantage of the individual repertoire of the patient’s immune system, targets CD4⁺ and CD8⁺ T cells alike, and thereby meets the need for a CD4⁺ T-cell response for effective antitumor immunity [54].

Viral vectors, such as vaccinia or adenoviruses, allow the efficient expression of the antigen coding gene in somatic cells. In animal models, recombinant MVA has been shown to elicit immune responses to its inserted antigens both in vitro and in vivo [13]. We conducted a phase I vaccination trial with MVA-hTyr to determine its feasibility and safety in patients with AJCC stage II disease [20] as adjuvant vaccine.

The application of MVA-hTyr was well tolerated by all 20 patients. The study medication caused few side effects, most of which were inflammatory and limited to the application site. No WHO grade III or IV toxicity occurred. One serious adverse event was reported. Patient 2 had a relapse of the melanoma at the excision site of the primary tumor after the second vaccination and was therefore withdrawn from the study. The relapse was assessed as not likely to be related to the study medication. The good tolerability of MVA is in line with earlier results obtained during the two-step vaccination protocol against smallpox in Germany [35] and also with results of recent trials with recombinant vaccinia viruses [8]. The data presented here demonstrate that MVA is strongly immunogenic. We found an increase in virus-specific CD8⁺ and CD4⁺ T-cell responses, as well as an increase in antiviral antibody titers in all patients tested. These findings are confirmed by recently published data, demonstrating the efficacy of vaccina virus in inducing long-lasting humoral and cellular immune responses [16, 18].

Testing for antibodies against tyrosinase did not reveal an increase in titer during vaccination in our patients. It has been shown that antibodies against tyrosinase detected in patients with vitiligo are heterogeneous [27]. The tyrosinase protein fragment used as target for antibody detection in this study completely covers three of four epitopes, known to elicit antityrosinase antibodies in the vitiligo patients. Therefore, the probability that we missed a vaccine-induced specific antibody response to tyrosinase is low.

The antityrosinase T-cell response was analyzed with different antigen formats mainly in ELISpot assays. A positive result in ELISpot assays on ex vivo blood lymphocytes has been correlated with previous antigen exposure [21]. ELISpot assays are therefore widely used for the monitoring of vaccination efficacy in clinical trials [44, 53, 58]. PBMCs of at least three different samples of every patient were tested for CD8⁺ T cells recognizing known tyrosinase peptides. To rule out the possibility that peptide-specific T cells were missed because they did not produce IFN-γ, we performed confirmatory analyses with HLA-A2 tetramers complexed with peptide hTyr_369–377. To overcome the restriction to known peptides, as well as to be able to detect not only CD8⁺ but also CD4⁺ T cells, autologous DCs expressing full-length tyrosinase were used as stimulators in ELISpot assays. First DCs infected with MVA-hTyr or with recombinant adenovirus carrying the tyrosinase gene were applied. Both resulted in the detection of strong preexisting antiviral reactivity, possibly covering rather weak antityrosinase responses. To avoid background reactivity against vector components, DCs transfected with tyrosinase mRNA were used in a second step. Indeed, transfection of either CMV pp65-mRNA or tyrosinase-mRNA caused no background reactivity and was equivalent to exogenous loading of APCs with peptides in its ability to induce IFN-γ spot formation [4]. Nevertheless, no tyrosinase-specific CD4⁺ and CD8⁺ T-cell response was seen ex vivo.

It has been reported that in some patients vaccinated with peptides of different MDA, peptide-specific T cells were detected only in the lymph nodes and were undetectable in the peripheral blood [60]. By using the in vitro amplification prior to testing [9] we increased the sensitivity for detection beyond that of ex vivo tests. The lack of increase in T-cell frequencies against tyrosinase that was detected in these assays suggests an incapability of our vaccination protocol to induce or even to boost a tyrosinase-specific T-cell response. This became especially obvious in patient 6, who was the only one to show a detectable preexisting T-cell response to two peptides of tyrosinase, which did not increase after vaccination.

This incapability to induce a measurable antityrosinase immune response can be explained in several ways. Vaccinia virus infection has been shown to impair the maturation of dendritic cells [15]. However, both mature and immature DCs are susceptible to infection with recombinant vaccinia viruses, and both express transgenes sufficiently [38]. The strong humoral and cellular response to MVA induced by vaccination with MVA-hTyr rules out a relevant impairment of antigen presentation and costimulation by MVA-hTyr with certainty.

Recent vaccination studies with recombinant viruses in mice have shown that preexisting antiviral CTLs impair the induction of a T-cell response to their transgene, especially if viral and transgene-encoded peptides share the same HLA restriction [47]. High-avidity T cells lyse infected cells more rapidly and at lower antigen densities than low-avidity T cells, as shown in a murine model [10]. Fast lysis of infected APCs by MVA-specific T cells might abrogate an immune response to tyrosinase. Beyond that, preexisting antiviral antibodies may counteract the induction of transgene-specific immune reactions. The strong humoral and cellular response to MVA found in all vaccinated patients may therefore represent one reason for the lack of induction of a tyrosinase-specific immune response. Prime-boost vaccinations using different vectors for gene delivery might help to circumvent the impact of a strong antivector immunity, and indeed some promising results have been achieved with nonhuman primates [45]. However, even with prime-boost vaccination the presence of neutralizing anti-MVA antibodies was able to impair the induction of an immune response against the transgene, as shown for the vaccination with the measles hemagglutinin [57].

Preexisting anti-MVA immunity in the patients in our study does not completely explain the failure to induce tyrosinase-specific responses, because some patients were either not vaccinated before or did not show a measurable pre-formed antiviral reactivity. Tyrosinase is a self-antigen not exclusively expressed in melanoma. It is therefore likely to induce tolerance. Whereas vaccination of HLA-A2–transgenic mice with MVA carrying the human tyrosinase gene led to the induction of a measurable specific immune response [13], vaccination with vaccinia carrying the mouse homologue of tyrosinase did not [40]. The use of xenogeneic homologues or the directed alteration of the amino acid sequence of MAA might therefore serve as a means to break tolerance. A promising approach to increasing immunogenicity by sequence alteration of a self-antigen is the in-frame fusion of a strong helper epitope to the homologous protein. This strategy has already been successfully applied in mice [49]. Another way to break tolerance is the use of dendritic cells as antigen delivery system, leading to specific T-cell responses against homologous MAA in mice [2] and in humans [29, 39, 53].

In summary, recombinant MVA is a safe delivery system for antigens. Although the virus is highly immunogenic, it is however not sufficient to break tolerance to tyrosinase in patients with stage II melanoma after repeated combined intradermal and subcutaneous vaccinations. Additional strategies, such as alternative routes of vaccine administration [51], prime-boost protocols [45], ex vivo infected/transfected DCs [19], or the in-frame fusion of strong helper epitopes to the antigens [49] should be included in further vaccination trials.

Abbreviations

AJCC

American Joint Committee on Cancer

APC

Antigen-presenting cell

BN

Bavarian Nordic

CEF

Chicken embryo fibroblast

CTL

Cytotoxic T lymphocyte

ELISA

Enzyme-linked immunosorbent assay

ELISpot

Enzyme-linked immunosorbent spot

IFN

Interferon

IL

Interleukin

MAA

Melanoma-associated antigen

MDA

Melanosomal differentiation antigen

MVA

Modified vaccinia virus Ankara

PBMC

Peripheral blood mononuclear cell

UPN

Unique patient number