Potent costimulation of human CD8 T cells by anti-4-1BB and anti-CD28 on synthetic artificial antigen presenting cells

Abstract

The in vitro generation of cytotoxic T lymphocytes (CTLs) for anticancer immunotherapy is a promising approach to take patient-specific therapy from the bench to the bedside. Two criteria must be met by protocols for the expansion of CTLs: high yield of functional cells and suitability for good manufacturing practice (GMP). The antigen presenting cells (APCs) used to expand the CTLs are the key to achieving both targets but they pose a challenge: Unspecific stimulation is not feasible because only memory T cells are expanded and not rare naïve CTL precursors; in addition, antigen-specific stimulation by cell-based APCs is cumbersome and problematic in a clinical setting. However, synthetic artificial APCs which can be loaded reproducibly with MHC-peptide monomers and antibodies specific for costimulatory molecules could resolve these problems. The purpose of this study was to investigate the potential of complex synthetic artificial APCs in triggering the costimulatory molecules CD28 and 4-1BB on the T cell. Anti-4-1BB antibodies were added to an established system of microbeads coated with MHC-peptide monomers and anti-CD28. Triggering via CD28 and 4-1BB resulted in strong costimulatory synergy. The quantitative ratio between these signals determined the outcome of the stimulation with optimal results when anti-4-1BB and anti-CD28 were applied in a 3:1 ratio. Functional CTLs of an effector memory subtype (CD45RA⁻ CCR7⁻) were generated in high numbers. We present a highly defined APC platform using off-the-shelf reagents for the convenient generation of large numbers of antigen-specific CTLs.

Introduction

CD8 T cells are key effectors in the battle against viral pathogens. In cancer immunology, they help to fight tumour cells. Many tumour-associated T-cell epitopes have been defined and this has enabled the design of highly specific anti-cancer regimens [34]. Due to the lack of functional tumour-specific cytotoxic T lymphocytes in many cancer patients, there is a need for the development of methods to prime and expand tumour-specific CTLs. Adoptive immunotherapy is an approach which involves in vitro priming and expansion of T lymphocytes with subsequent in vivo infusion of expanded CTLs. Dendritic cells have been used as APCs for the in vitro priming of CTLs [3]. This method involves the laborious differentiation of autologous peripheral blood monocytes or CD34⁺ hematopoietic precursors. A problem of this strategy is the limited number of dendritic cells that can be generated. Furthermore, only dendritic cells which are fully mature can prime T cells; immature dendritic cells induce tolerance [18].

The advent of artificial APCs has raised hopes that these shortcomings might soon be overcome. Many different artificial APCs have been designed—either cell-based or synthetic artificial APCs. First, artificial APCs were developed for the unspecific expansion of T cells. These approaches took advantage of anti-CD3 and anti-CD28 antibodies which were added to T cells [27] and used to decorate cell lines such as K562 [37] or microbeads [38]. However, these methods are unsuitable for anti-cancer immunotherapy when rare precursor T cells need to be primed and expanded in an antigen-specific fashion. Thus, methods for antigen-specific T cell priming and expansion have been established. Various examples of cell-based strategies have been presented: they include HLA-A2-negative B cells loaded with HLA-A2 MHC-peptide monomers [30], mouse fibroblasts retrovirally transduced with an HLA–peptide complex plus the accessory proteins B7.1, ICAM-1, and LFA-3 [16] and insect cells transfected with single-chain HLA, CD54 and CD80 [11]. These protocols have shown that the generation of antigen-specific T cells by employing artificial APCs is feasible. However, the use of such artificial APCs in a GMP setting is problematic because cell-based reagents are difficult to implement in GMP protocols.

In contrast, synthetic artificial APCs, which are made from a limited number of “off-the-shelf” components, could be more suitable for the clinic. Previous studies have established protocols that involve microbeads loaded with MHC-peptide molecules in monomeric [42] or Ig-coupled dimeric forms [24], and anti-CD28. Clearly, the engineering of artificial APCs is not only a qualitative problem (in terms of the receptors on the T cell to be triggered) but also a quantitative challenge since APCs in vivo display a carefully controlled expression pattern of costimulatory molecules. This issue was neglected during the design of cell-based artificial APCs when cell lines were engineered to express different costimulatory molecules (see above).

To address these issues, we hypothesized that the use of further stimulatory antibodies in addition to anti-CD28 could render synthetic artificial APCs more powerful, especially if the stimulatory signals were applied in an optimised ratio. To demonstrate this, we took advantage of a strictly controlled system of microbeads loaded with defined amounts of MHC-peptide monomers and anti-CD28 antibodies that was previously used to prime and expand CTLs [31, 42]. To assess possible synergistic effects between costimulatory antibodies, anti-4-1BB was added to the system in the present study. This antibody triggers the tumour necrosis factor (TNF) receptor family member 4-1BB (CD137) [14]. Its ligand 4-1BBL is expressed by APCs such as activated B cells and dendritic cells [9]. The costimulatory molecule 4-1BB is upregulated within 24 h after T-cell activation [7] and signals independently from CD28 [12]. 4-1BB signalling primarily induces CD8 T-cell proliferation in vitro and leads to the amplification of cytotoxic T-cell responses in vivo [33]. 4-1BB was characterized as a T-cell survival signal [35] which may be due to its anti-apoptotic effects [17]. Several studies have shown anti-tumour and antiviral effects of 4-1BB costimulation [6, 20, 21, 36]. 4-1BBL was demonstrated to be a valuable costimulatory factor for the unspecific expansion of T cells via anti-CD3 and anti-CD28 [19, 43].

In this study, we found a synergy between anti-CD28 and anti-4-1BB antibodies during T-cell priming and expansion that was dependent on the antibody ratio; we report the highest percentage of peptide-specific CTLs generated by antigen-specific in vitro priming so far. The CTLs displayed an effector memory phenotype and were functional, thereby proving their value for therapeutic approaches. An effective protocol for the in vitro generation of CTLs is provided and the importance of highly defined, synthetic artificial APCs is highlighted.

Materials and methods

Peptides and MHC-peptide monomers

The Melan-A-derived peptide ELAGIGILTV [40] was synthesized using standard Fmoc/tBu chemistry [32]. Biotinylated recombinant HLA-A*0201 molecules and fluorescent MHC-tetramers were produced as described previously [1]. Briefly, fluorescent tetramers were generated by incubating biotinylated HLA monomers with streptavidin-PE or streptavidin-APC (Molecular Probes, Leiden, The Netherlands) at a 4:1 molar ratio.

Biotinylated antibodies

The antibodies mouse IgG2a anti-human CD8 Ab OKT-8, mouse IgG2a anti-human CD28 Ab 9.3, mouse IgG1 anti-human 4-1BB Ab 4B4-1 (Becton Dickinson Biosciences (BD), Heidelberg, Germany) or [14] were biotinylated using sulfo-N-hydroxysuccinimidobiotin (Perbio Science, Bonn, Germany) as recommended by the manufacturer.

Synthetic artificial APCs

About 5.6 μm-diameter streptavidin-coated polystyrene particles (BangsLabs, Fishers, IL, USA) were resuspended at 4 × 10⁶ microbeads per ml in PBE (PBS/BSA/EDTA) was PBS (phosphate-buffered saline) (BioWhittaker, Verviers, Belgium) containing 0.5% of bovine serum albumin (Sigma, Aldrich) and 2 mM EDTA (Roth, Karlsruhe, Germany), containing 200 pM biotinylated MHC-peptide monomer and a total antibody concentration of 20 nM and incubated at room temperature for 30 min. After washing, the synthetic artificial APCs were stored at 4°C prior to use.

In vitro priming and expansion of human CD8 T cells

Fresh HLA-A*02⁺ buffy coats or leukapheresis products were used to isolate PBMCs by standard gradient separation. CD8 T cells were MACS-enriched by biotinylated OKT-8 antibody and Streptavidin-Microbeads (Miltenyi-Biotec, Bergisch-Gladbach, Germany). Stimulations were initiated in 96-well plates with 1 × 10⁶ responder cells plus 2 × 10⁵ microbeads in 250 μl of T-cell medium [42] complemented with 5 ng/ml human IL-12 p70 (PromoKine, Heidelberg, Germany). After 3–4 days of incubation at 37°C, fresh medium with 80 U/ml human IL-2 (Chiron, Emeryville, CA, USA) was added and cells were incubated for another 3–4 days. This stimulation cycle was performed three/four times.

CFSE-based proliferation assay and cloning

To evaluate the proliferative response to the MelanA antigen, 70 × 10⁶ CD8⁺ T cells were labeled with 5 μM CFSE (5/6-carboxyfluorescein diacetate, succinimidyl ester, Molecular Probes, Leiden, The Netherlands) and incubated for 10 min at 37°C in the dark. To stop labeling, an equal volume of T-cell medium complemented with 20% heat-inactivated fetal bovine serum (FCS) (PAA, Cölbe, Germany) was added for another 20 min. Cells were then washed in T-cell medium four times. About 10⁶ CFSE-labeled cells were seeded into individual wells of a 96-well culture plate, synthetic artificial APCs were added as indicated. CD8⁺ T cells were incubated at 37°C, 5% CO₂. Antigen-stimulated CFSE-labeled CD8⁺ T cells were analyzed by flow cytometry on a four-color FACSCalibur cytometer (BD).

Flow cytometry

Tetrameric analyses were performed with tetramer-PE/APC plus antibody CD8 PerCP clone SK1 (BD). Cells were incubated with the antibody at 4°C for 20 min in the dark, followed by 30 min incubation with fluorescent MHC tetramers at the same conditions. After washing, cells were analyzed by flow cytometry on a four-color FACSCalibur cytometer (BD). Total specific cell numbers per sample could be calculated after FACS analysis: (specific cells counted) × (microbeads added)/(microbeads counted).

For phenotyping CCR7 staining was performed using rat hybridoma supernatant 3D12 (kindly provided by R. Förster, Anova) and donkey anti-rat (Fab’)₂-PE fragments (Jackson ImmunoResearch, West Grove, PA, USA) or hCCR7 PE (R&D, Wiesbaden, Germany). After blocking with 10% heat-inactivated mouse serum (CC pro, Neustadt, Germany), cells were further stained with CD45RA-FITC (BD) and tetramer-APC.

Analysis of surface costimulatory molecules, CD137 (4-1BB) and CD28 staining was performed using anti-human CD137-PE (BD) and anti-human CD28-FITC (Immunotools, Friesoythe).

Generation of T-cell lines

Sorting was done with a FACSVantage cell sorter. Sorted tetramer-positive cells were expanded by PHA-L, IL-2, and feeder cells (irradiated LG2-EBV and irradiated allogeneic PBMCs) as described before [42].

T-cell assays

Intracellular IFN-γ staining was done as previously described [23]. Briefly, T cells were stimulated with 10 μg/ml peptide for 6 h in the presence of Golgi-Stop (BD). Cells were analyzed using a Cytofix/Cytoperm Plus kit (BD) and an IFN γ-PE or IFN γ-FITC antibody (BD). After staining, cells were analyzed on a four-colour FACSCalibur (BD).

Cytotoxicity was tested in a standard 4 h ⁵¹Cr release assay using 3,000 target cells per well. Percentage of specific lysis was calculated as follows: (experimental release − spontaneous release)/(total release − spontaneous release) × 100

Results

Synergy between anti-CD28 and anti-4-1BB in CD8 T-cell costimulation is most pronounced when applied in a specific ratio

To assess the stimulative capacity of anti-CD28, anti-4-1BB and a potential synergy of the antibodies, beads were coated with MHC-peptide-monomer and loaded either with a single antibody or with both antibodies at different ratios.

First, the proportion of antigen-specific CTLs prior to stimulation was determined by flow cytometry (Fig. 2a): in an ideal setting with a purity of 100% CD8⁺ enriched T cells, 0.1% antigen-specific CTLs correspond to a total number of 1,000 antigen-specific CTLs per well (10⁶ CD8-enriched cells were used per well).

Fig. 2.

Highly elevated levels of antigen-specific effector memory CTLs after stimulation. a PBMCs of a healthy HLA-A2-positive donor were stained ex vivo with CD8-PerCP and tetramer to detect the frequency of antigen-specific T cells before stimulation. Due to unspecific tetramer binding to B-cells, only CD8⁺ cells are shown. b CD8-enriched T cells of the same donor were stimulated four times in vitro with A*0201/ELAGIGILTV-coated microbeads loaded with anti-CD28 and anti-4-1BB in a 1:3 ratio. Positive T cells were detected by tetramer staining. c T cells were stained with CD8-PerCP, CD45RA-FITC, CCR7 (rat hybridoma supernatant and donkey anti-rat-PE) and tetramer-APC to detect the percentage of tetramer-positive cells and their phenotype. The percentage of tetramer⁺ CD8⁺ lymphocytes per quadrant is indicated. Tetramer-positive cells are displayed in black (highlighted), tetramer-negative cells in grey. Assays were performed ex vivo (c) and after four rounds of stimulation (d)

After three rounds of stimulation, moderate T-cell priming and expansion were detected with a single costimulatory antibody (Fig. 1a). However, combining both antibodies resulted in a higher percentage of tetramer-positive cells. A 3:1 ratio between anti-4-1BB and anti-CD28 increased the percentage of tetramer-positive cells to values of up to 48% after three rounds of stimulation. In the experiment shown in Fig. 1a, about 1,47,000 tetramer-positive CTLs were detected using the optimal 3:1 ratio of the costimulatory antibodies compared with an average of 21,000 CTLs when using anti-CD28 only.

Fig. 1.

Anti-CD28 and anti-4-1BB on MHC-coated microbeads synergize in T-cell priming and expansion. Quintuplicates of CD8-enriched T cells of a healthy HLA-A2-positive donor were stimulated three times (a) or four times (b) in vitro with A*0201/ELAGIGILTV-coated microbeads loaded with either anti-CD28, anti-4-1BB or a combination of both. Positive T cells were detected by tetramer staining. Repeated stimulation of CD8-enriched T cells of different donors demonstrated that on average a fivefold higher percentage of tetramer-specific T cells can be obtained by costimulation with anti-4-1BB/anti-CD28 in a 3:1 ratio if compared to anti-CD28 alone (inset Fig. 1a)

In vitro stimulation of CD8⁺ T cells from six other healthy donors confirmed that a ratio of 3:1 of anti-4-1BB to anti-CD28 was superior to anti-CD28 alone (Fig. 1a inset): on average, a fivefold higher percentage of specific CTLs was obtained. When either anti-CD28 or anti-4-1BB were replaced by an isotype control, no synergistic effect was observed (data not shown). Thus, any effects due to antibody dilution could be excluded.

Elevated levels of antigen-specific CTLs after four rounds of stimulation

Usually three rounds of stimulation are performed during in vitro priming. Here, we demonstrate that a further round of stimulation can result in a much higher percentage of antigen-specific cells. For the optimal ratio between anti-CD28 and anti-4-1BB, values of up to 72% antigen-specific T cells were monitored (Fig. 1b).

To our knowledge this is the highest value of antigen-specific cells observed for in vitro priming experiments.

Priming of naïve T cells by synthetic artificial APCs results in the generation of CTLs of the effector memory phenotype

For phenotype analysis, cells were stained with tetramer and appropriate antibodies. In one stimulation process, 62% tetramer-positive T cells were detected after four rounds of stimulation (Fig. 2b). Of these tetramer-positive cells, 88% were effector memory (CCR7⁻ CD45RA⁻), 9% central memory (CCR7⁺ CD45RA⁻), 2% effector (CCR7⁻ CD45RA⁺) and 1% naïve T cells (CCR7⁺ CD45RA⁺) (Fig. 2d). These results were confirmed after stimulation and after phenotyping T cells of a second donor, where 80% of effector memory, 10% central memory, 9% effector and about 1% naïve T cells were obtained. To compare the phenotype of these tetramer⁺ specific cells with tetramer⁺ cells prior to stimulation we also analyzed PBMCs of the same donors ex vivo. One exemplary result is shown in Fig. 2c. About 70% of tetramer⁺ cells analyzed ex vivo were of a naïve phenotype (CCR7⁺ CD45RA⁺), as expected from previous reports [44].

The generated CTLs are functional and exert effector functions in a specific fashion

For functional analysis, tetramer-positive T cells generated by the optimal antibody ratio were sorted and three cell lines were established.

Functionality of the T cells was demonstrated by intracellular IFN-γ staining and chromium release assay. After stimulation with the Melan-A-expressing HLA-A*02⁺ melanoma cell line MeWo, all T-cell lines produced IFN-γ (one representative experiment is shown in Fig. 3a), while irrelevant targets did not stimulate IFN-γ production (data not shown). INF-γ production was also observed after stimulation of the T cell lines with either ELAIGILTV-loaded T2 cells (Fig. 3b) or the Melan A expressing HLA-A*02⁺ melanoma cell line Mel-CG1 (Fig. 3c).(On average, 30–50% of tetramer-specific CTLs produce INF-γ((Fig. 3a–c). These results correspond well with data of generated virus specific CTLs (D. Rudolf, unpublished) and emphasize our notion that tetramer staining indicates specificity but not functionality of T cells. Furthermore, the same T-cell line lysed the Melan-A-positive targets specifically and efficiently (MeWo, Mel-CG-1 and ELAGIGILTV-loaded T2 cells, but there was no lysis of control targets (BV-173 cells and T2 cells loaded with an HIV-peptide) (Fig. 3d). To exclude unspecific lysis of the melanoma cell lines, we tested T-cell lines generated from the sorted tetramer-negative cells. Neither ELAGIGILTV-loaded T2 cells nor MeWo cells were lysed by the control cell lines (data not shown). This indicates that the tetramer-positive CTLs killed the melanoma cells in an antigen-specific fashion.

Fig. 3.

Tetramer-positive cells are fully functional cytotoxic T lymphocytes. Intracellular IFN-γ staining of sorted tetramer-positive T cells stimulated with a HLA-A*02-positive Melan-A-positive MeWo cells. b ELAGIGLTV-loaded T2 cells, c HLA-A*02-positive Melan-A-positive Mel-CG1 cells. The percentage of IFN-γ producing CD8-positive T cells among tetramer-positive cells is indicated. d The killing activity of the same T-cell lines was tested against ELAGIGILTV-loaded TAP-deficient T2 cells (filled square), the HLA-A*02-positive Melan-A-positive melanoma cell lines MeWo (filled diamond) and Mel-CG-1 (filled triangle). ILKEPVHGV-loaded T2-cells (HIV-peptide, open square) and the HLA-A*02-positive leukemia cell line BV-173 (open triangle) were used as control. One of three independent experiments with comparable results is shown

CD8, CD28, and 4-1BB surface marker expression during stimulation of CD8⁺ T cells with aAPCs

To assess cell division and surface expression of the costimulatory molecules CD28 and 4-1BB during in vitro stimulation, naive CD8⁺ T cells were labeled with CFSE and then stimulated with A*0201/ELAGIGILTV-coated microbeads loaded with anti-CD28 and anti-4-1BB in a 1:3 ratio. After three stimulations, cells were stained with tetramer and appropriate antibodies.

The CFSE labeling indicates that within the CD8⁺ T cell population, tetramer⁺ T cells proliferated almost exclusively during stimulation. The activation was accompanied by a slight down regulation of the CD8 surface molecule in part of the T cells (Fig. 4a). Expression of 4-1BB was low in CD8⁺ T cells before stimulation, as analyzed by ex vivo staining. After three rounds of stimulation, 4-1BB was expressed again at low levels in tetramer-positive cells while tetramer-negative cells displayed higher 4-1BB expression (Fig. 4b–c): only 10% of tetramer-positive cells expressed 4-1BB compared to 90% of tetramer-negative cells expressing 4-1BB on their surface. The expression of the costimulating molecule CD28 was determined after in vitro stimulation: 44% of CD8⁺ T cells expressed CD28 after stimulation; and 63% of this CD28⁺ population were tetramer⁺ T cells (Fig. 4d–e).

Fig. 4.

Expression of surface markers CD8, 4-1BB and CD28 during stimulation. CD8 enriched T cells of a healthy HLA-A2-positive donor were labeled with CFSE and stimulated three times with A*0201/ELAGIGILTV-coated microbeads and an optimized ratio of costimulatory antibodies. Cell division was assessed at various time points by monitoring the CFSE intensity of these cultures, followed by staining with anti-CD8 and tetramer staining. a only tetramer⁺ CD8⁺ lymphocytes proliferate during stimulation with artificial APCs. Proliferation is accompanied by a slight decrease of surface CD8. 4-1BB expression is generally low prior to stimulation (b). After three stimulations, 4-1BB surface expression is low in tetramer⁺ T cells compared to tetramer⁻ T cells (c). In contrast, CD28 expression appeared high in tetramer⁺ CD8⁺ lymphocytes after stimulation (e) but significant expression was also determined ex vivo (d). The percentage of PBMCs per quadrant is indicated in all panels. Tetramer-positive cells are displayed in black (highlighted), tetramer-negative cells in grey

In summary, the data reveal that when used in a defined ratio, anti-CD28 and anti-4-1BB boost T-cell priming and expansion by synthetic artificial APCs. As only a specific ratio of the antibodies resulted in optimal stimulation, this study stresses the importance of defined artificial APCs for in vitro T-cell priming and expansion.

Discussion

The main goal of this study was demonstrate that synthetic artificial APC could be used as a tool in adoptive cancer immunotherapy if the stimulatory capacity of this APC platform is improved. A parallel comparison recently demonstrated that synthetic bead-based aAPC are superior to dendritic cells [42]. Here, we introduce an additional costimulating compound and also consider the ratio of costimulatory triggers and its effect on the outcome of stimulation, which is an issue that has been neglected until now.

Biotinylated MHC-peptide complexes and biotinylated antibodies were coupled onto streptavidin-coated microbeads. We used a well-described CTL epitope for the study, namely the modified Melan-A-derived peptide ELAGIGILTV [40] that has already been used as a target in melanoma immunotherapy. T cell stimulation resulted in greatly increased populations of antigen-specific CTLs upon stimulation with anti-CD28 and anti-4-1BB. The peak of 48% positive CTLs (1,47,000 cells in absolute numbers) after three rounds of stimulation was reached upon stimulation with artificial APCs coated with anti-4-1BB and anti-CD28 in a 3:1 ratio. Thus, approximately 10 ml of peripheral blood gave rise to nearly 1,50,000 antigen-specific CTLs.

The large majority of the generated CTLs displayed an effector memory phenotype which had been observed previously in other in vitro priming experiments [31, 42]. They can be distinguished from central memory T cells by their lack of the CCR7 chemokine receptor. Effector memory T cells were shown to migrate to peripheral tissues and to exert an immediate effector function, thereby providing protective memory [8, 28, 29]. The effector function of the generated CTLs was assessed by two functional assays: The CTLs were shown to be capable of antigen-specific IFN-γ production and target cell lysis. Unspecific lysis of cells, which would contradict usage in adoptive immunotherapy, was not detectable.

On average, 30–50% of tetramer⁺ T cells obtained after three rounds of stimulation were able to secrete IFN-γ. Other groups achieved higher amounts of IFN-γ secreting cells, specific for the same modified Melan A peptide ELAGIGLTV, only after enrichment using an IFN-γ antibody [25] or after cloning. Bulk cultures or T cell lines usually encompass different subsets with different effector functions [26], including even dysfunctional cells [41]. The risk of inducing dysfunctional T cells during strong and rapid expansion has been mentioned before [22].

CD28 is critical for initial T cell expansion, whereas 4-1BB/4-1BBL signalling impacts T cell numbers much later in the response and is required for the survival and/or responsiveness of the memory CD8 T cell pool. Upon engagement with 4-1BB ligand, 4-1BB can offer a CD28-independent costimulatory signal leading to CD4/CD8 T cell expansion, cytokine production, development of CTL effector function, and prevention of activated induced cell death [12]. 4-1BB is hardly detectable on the surface of freshly isolated human peripheral blood T cells but the expression of 4-1BB is inducible on human T cells. The 4-1BB molecule has been shown to be transiently expressed after TCR engagement in a precise and narrow window of time [7, 9]. In vitro the expression peaks at ∼48 h post-activation, is down-regulated and subsequently remains at a low, almost constant level [15]. We have been unable to detect significant 4-1BB expression on the tetramer-positive T cells due to the low level and transient nature of its expression. Members of the TNFR family, including 4-1BB are known to influence cell survival by activating the NF-κB pathway [2, 5], which in turn can lead to upregulation of Bcl-x_L as well as cellular inhibitors of apoptosis 1 and 2.

CD28 as well as members of the TNFR family influence Bcl-x_L expression [4, 10, 13, 17]. A variable percentage of unstimulated thymocytes are CD28⁺ but this is found in high density only on the CD3^{+ (bright)} CD4/CD8 single positive population. In our experiments, 63% of activated tetramer-specific CD8⁺ T cells expressed CD28 after three weeks of stimulation. CD28 expression increases upon cell activation after 12–24 h and persists for at least 6 days [39]. CD28 appears to be responsible for a first spate of survival signals and this can then be sustained by inducible costimulatory pathways such as 4-1BB. In the light of these findings, to perform optimal T cell priming we used artificial APCs coated with anti-4-1BB and anti-CD28 in a 3:1 ratio. We were able to demonstrate that this artificial system is highly effective for in vitro priming.

Adoptive immunotherapy of cancer relies on the priming and expansion of rare CTL precursors that are usually naïve CD8⁺ T cells. Obviously, these cells would be lost in a setting of unspecific T-cell expansion via anti-CD3 signals where memory T cells have a growth advantage. Hence, antigen-specific T-cell priming and expansion is needed for adoptive immunotherapy of cancer. Until now, most artificial APC approaches for antigen-specific T-cell stimulation have relied on genetically engineered cell lines. Clearly, this is a suboptimal solution in clinical immunology where standardised reagents are needed. In this context, synthetic artificial APCs are preferable for three reasons:

First, as no living cells are employed, synthetic artificial APCs are true “off-the-shelf” reagents; culturing APCs is more cumbersome than using synthetic agents. Synthetic artificial APCs can be produced ad libitum within the shortest time: biotinylated antibodies, MHC-peptide monomers and microbeads are commercially available and can be stored for long periods of time. They can be tailored to suit since all stimulatory parameters can be adjusted. In addition, synthetic artificial APCs are convenient. Prior to in vitro stimulation, sufficient APCs for all three rounds of stimulation are prepared and stored at 4°C. Each week, an aliquot is transferred to the stimulation experiment and sufficient numbers of APCs from one batch are therefore available for the entire stimulation procedure.

Second, the culture of cell-based artificial APC lines requires repeated screenings to ensure the quality of the lines. This is an obstacle in establishing immunotherapy as a routine procedure. An APC system that requires no maintenance operations will support the use of immunotherapy for a greater number of patients. For patients with progressive disease in particular it is vital to provide an APC system that is both readily available and reliable.

Third, cellular APCs are defined to a lesser extent than synthetic approaches. In our view, this is a drawback of cell-based APCs, such as dendritic cells or genetically engineered cell lines. It is not only cumbersome and expensive to generate dendritic cells, it is also difficult to produce fully-mature dendritic cells reproducibly that are necessary for in vitro priming—especially for different donors. Cell-based artificial APCs usually express the molecules involved in the T-cell priming process as transgenes. Hence, the number of stimulatory molecules at the cell surface is unknown and, more important, it is not possible to tune the expression of costimulatory molecules in order to provide the optimal ratio for stimulation.

Another issue is the avidity of in vitro generated CTLs. It was shown that only a low concentration of MHC-peptide monomers on artificial APCs resulted in high-avidity T cells [40]. Conversely, a high concentration of MHC-peptide monomers resulted in low-avidity T cells. This finding is often not taken into consideration when cell-based artificial APCs transfected with HLA molecules are used [16, 23]. The amount of “signal 1” that these systems provide is undefined and may vary according to culture conditions. Interestingly, just the combination of two costimulatory signals resulted in an average of 72% positive CTLs after four stimulations, which to our knowledge is the highest value observed in an antigen-specific in vitro priming experiment. Hence, synthetic artificial APCs are not only convenient and versatile but also powerful tools for T-cell priming and expansion.

In sum, the introduction of complex yet defined synthetic APCs should allow the convenient generation of large numbers of high-avidity CTLs by an optimized ratio of MHC-peptide complexes and costimulatory triggers for both therapeutic and experimental use.