Mature But Not Immature Fas Ligand (CD95L)-Transduced Human Monocyte-Derived Dendritic Cells Are Protected from Fas-Mediated Apoptosis and Can Be Used as Killer APC¹

Dendritic cells (DC)³ are a heterogeneous population of bone marrow-derived cells present in most peripheral tissues that are able to capture and present Ags to the cells of the adaptive immune system (1). Such antigenic presentation to T cells can lead to two opposite outcomes: potent activation (immunogenicity); or inhibition (tolerance) of effector immune functions (2, 3). Recent progress in understanding of the physiological function of DC and technological advances in generating large numbers of DC from various progenitors in vitro have led to the development of many DC-based vaccines as a potent strategy to initiate protective immune responses against tumors and infectious pathogens. In contrast, novel strategies were developed to enhance the capacity of DC to induce immunological tolerance. These strategies include pretreatment of DC with IL-10 (4), UV irradiation (5), use of DC in an immature state (6), or transduction of DC with immunoregulatory molecules such as IL-10 (7), TGF-β (8), and CTLA-4 (9).

One central mechanism of immune privilege is use of the Fas/Fas ligand (FasL)-mediated apoptosis to delete invading T cells at immune privilege sites, and constitutive production of FasL has been demonstrated by Sertoli cells of the testes and the retinal cells of the eye (10). In addition, expression of FasL has been observed in different tumor cells suggesting that FasL/Fas-mediated apoptosis might be a critical mechanism by which tumors evade the immune response as proposed in the tumor counterattack model (11). With regard to these observations, studies were performed to modulate the interaction of APCs and T cells by triggering Fas apoptosis signaling, which should result in deletion of activated T cells as proposed for immune privilege tissues.

Ag-specific elimination of activated T cells by FasL-expressing APC has been demonstrated by us and other investigators in different in vitro and in vivo experimental models using murine cells or cell lines as APC. These encouraging results indicated the therapeutic potential of FasL-expressing killer APC as immunoregulatory cells for the treatment of allograft rejection (12, 13, 14, 15, 16), autoimmune disease (17, 18, 19, 20, 21), and chronic infections (22, 23). However, it has not been determined whether human DC can be efficiently transduced by FasL and whether these cells are capable to eliminate Fas⁺ target cells. In contrast to murine cells or cell lines, transduction of human monocyte-derived macrophages and DC is difficult, and high transduction rates could be achieved only with viral gene transfer strategies (24). In addition, in previous studies, FasL-expressing APC were generated from mice deficient in Fas-mediated apoptosis to prevent self-destruction of transduced FasL-expressing APC. Fas-mediated self-destruction also hampers the propagation of FasL-encoding viral vectors in Fas⁺ host cells (25). Therefore, experiments to generate FasL-expressing APC from primary human cells are still lacking. Because several studies demonstrated that FasL is highly conserved within different species and that murine FasL induces apoptosis in human Fas⁺ cells (18), we investigated in the present study whether primary human monocyte-derived immature DC (iDC) and mature DC (mDC) could be transduced with murine FasL using an inducible Cre/Loxp adenovirus-based gene transfer system to generate killer APC.

PBMC were isolated from leukapheresis concentrates of healthy donors, and monocytes were separated by countercurrent elutriation (26). To induce the in vitro differentiation of monocytes to DC, monocytes were cultured for 7 days in serum-free CellGro culture medium (CellGenix, Freiburg, Germany) in the presence of 500 U/ml IL-4 (Promocell, Heidelberg, Germany) and 500 U/ml GM-CSF (Leukomax; Essex, Munich, Germany). To induce maturation of DC, cells were additionally stimulated with IL-1β (10 ng/ml), TNF-α (10 ng/ml), IL-6 (1000 U/ml), all from Promocell (Heidelberg, Germany) and PGE₂ (1 μg/ml; Minprostin E₂; Pharmacia & Upjohn, Erlangen, Germany) for an additional 2 days as described previously (27). To confirm the phenotypes of iDC and mDC, expression of the markers CD83, CD80, CD86, and HLA-DR was determined in cultured DC by FACS analysis.

T cells were separated by countercurrent elutriation at a flow rate of 52 ml/min in HBSS and 6% autologous plasma. Cells were frozen immediately in RPMI culture medium containing 10% DMSO (Sigma-Aldrich, Steinheim, Germany) and 40% heat-inactivated (56°C for 30 min) autologous plasma. For additional experiments, T cells were rapidly thawed at 37°C, washed with PBS (Life Technologies Invitrogen, Paisley, U.K.), and resuspended in RPMI 1640 (BioWhittaker Europe, Verviers, Belgium) supplemented with 200 mM l-glutamine, 100 mM sodium pyruvate, nonessential amino acids, minimal essential medium vitamins (all from Life Technologies), penicillin/streptomycin (50 U/50 μg/ml; PAA Laboratories, Linz, Austria) and 50 μM 2-ME (Amresco, Solon, OH). For polyclonal activation, aliquots of T cells were stimulated with PHA (1 μg/ml) for 5 days.

Jurkat T cells (purchased from DSMZ, Braunschweig, Germany) were cultured in RPMI 1640 (PAN Biotech, Aidenbach, Germany) supplemented with 10% FCS, 100 U/ml penicillin, 0.1 mg/ml streptomycin (all from PAA), 1 mM HEPES, and 1 μg/ml amphotericin B (both from Sigma-Aldrich, Deisenhof, Germany).

To prevent Fas/FasL-mediated apoptosis during virus propagation in Fas⁺ 293 cells and to facilitate inducible transgene expression, recombinant adenoviruses encoding murine full-length FasL were generated using the Cre/Loxp-inducible system. AdLoxpFasL was generated by separating the FasL gene from the modified chicken β-actin promoter with the CMV immediate-early enhancer by a Loxp-flanked stuffer segment as previously published. The Cre recombinase protein encoded in AxCANCre has been demonstrated to excise the stuffer and allow functional reconstitution of the expression cassette at the deleted E1 site of the recombinant adenovirus vector leading to FasL expression in cells simultaneously transduced with AxCANCre and AdLoxpFasL (28, 29). A recombinant adenovirus encoding EGFP (AdEGFP) was used as a control vector. Viruses were propagated in HEK 293 cells (Clontech, Heidelberg, Germany) and enriched by ultracentrifugation as described (30).

For transduction, iDC and mDC were used 9 days after initiation of cultures. According to a previously published protocol (31), DC were incubated at a concentration of 2 × 10⁶/ml in serum-free CellGro culture medium for 90 min with or without the different recombinant adenoviruses. Afterward, cells were resuspended at a concentration of 1 × 10⁶/ml by addition of fresh culture medium containing 500 U/ml IL-4 as well as 500 U/ml GM-CSF. A multiplicity of infection (MOI) of 200 was determined as optimal and used throughout additional experiments for single transductions, and a MOI of 100 for each vector was used in double-transduction experiments. On day 12 (3 days after transduction), untreated and transduced DC were used for additional experiments. For time course and dose dependency experiments, cells were transduced with different MOIs as indicated and analyzed on day 1, 2, or 3 after transduction.

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions involving an additional DNA digestion step for 30 min (RNase-free DNase Set; Qiagen). Total RNA was used to synthesize cDNA using the 1st Strand cDNA Synthesis Kit including avian myeloblastosis virus for reverse transcription (Roche Diagnostics, Mannheim, Germany). RT-PCR for the murine FasL was performed using the forward primer 5′-AGGAATGTATACGCTCTTCC-3′ and the reverse primer 5′-CCTCATATAGACCTTGTGGT-3′ (product 369 bp). Primers for amplification of 18s were forward 5′-TCAAGAACGAAAGTCGGAG-3′ and reverse 5′-GGACATCTAAGGGCATCACA-3′ (product 488 bp). RT-PCR was performed in a 20-μl reaction volume containing the Taq PCR Master Mix Kit (Qiagen), 10 pmol from each primer of a primer pair, and 2 μl of cDNA. All samples were incubated for 20 cycles (denaturation for 30 s at 94°C, annealing for 30 s at 55°C, and elongation for 30 s at 72°C) using a thermocycler (GeneAmp PCR System 9700; PE Applied Biosystems, Foster City, CA). RT-PCR products were separated on a 1% agarose gel (FCM Bioproducts, Rockland, ME) and visualized by GelStar staining (BioWhittaker Molecular, Rockland, ME).

After two washings with PBS containing 10% FCS (PAA), cells were stained for murine FasL on the surface (PE-conjugated, clone Kay-10) or isotype control PE-conjugated mouse anti-mouse IgG2b (both purchased from BD PharMingen, San Diego, CA). After incubation for 45 min on ice in the dark followed by two washing steps, cells were fixed in PBS containing 0,1% paraformaldehyde (Sigma-Aldrich), and cells were analyzed using a EPICSXL-MCL (Coulter Electronics, Miami, FL). For detection of endogenous expression of human Fas and human FasL, DC were stained with PE-conjugated anti-Fas (clone VI C-64), biotin-conjugated anti-FasL (clone NOK-1) in combination with PE-conjugated streptavidin, or isotype control Abs (PE-conjugated anti-mouse IgG1, all purchased from BD PharMingen).

For detection of apoptosis, cells were washed twice with ice cold PBS and stained simultaneously with FITC-conjugated annexin V and propidium iodide (PI) according to the manufacturer’s instructions (both from BD PharMingen) for 20 min on ice in the dark with a binding buffer containing 10 mM HEPES-NaOH, 140 mM NaCl, and 2.5 mM CaCl₂ (all from Sigma-Aldrich). Within the next hour, cells were analyzed for apoptosis. Total numbers of apoptotic cells were determined by calculation of annexin V⁺ and PI⁻ cells (reflecting early apoptosis) together with annexin V⁺ and PI⁺ cells (reflecting late apoptosis/secondary necrosis). Analysis of data was performed using the software WinMDI (version 2.8; http://facs.scripps.edu).

To determine FasL-mediated cytotoxicity, cocultures of Jurkat T cells or activated and nonactivated primary T cells were established with transduced DC at different E:T ratios. Cytotoxicity was detected either by FACS using annexin V and PI staining as described, or using the JAM assay as previously published (32, 33).

To label the Fas⁺ Jurkat T cells as targets for the JAM assay, 5 × 10⁵/ml cells were incubated overnight with 2,5 μCi/ml [methyl-³H]thymidine (Amersham Pharmacia, Erlangen, Germany). After two washing steps using PBS, 4 × 10⁴ labeled Jurkat T cells were cocultured with 10⁴ FasL-expressing DC or control DC. As a positive control, 4 × 10⁴ labeled Jurkat T cells were incubated with an activating anti-Fas Ab (0.5 μg/ml, clone CH11; Upstate, Charlottesville, VA). To determine dose dependency, different amounts of mDC transduced with AxCANCre and AdLoxpFasL were incubated with 5 × 10⁴ labeled Jurkat cells. To investigate whether cytotoxicity of FasL-expressing DC is cell contact dependent, additional Transwell experiments were performed, in which 4 × 10⁴ labeled Jurkat cells were placed in 96-well microtiter plates, and Transwells (Nunc, Roskilde, Denmark) containing 10⁴ FasL-expressing DC were added. After 5 h, cells were lysed, and DNA was transferred onto glass fiber filters (Printed Filtermat B; Wallac Oy, Turku, Finland) using a Vacusafe IH-280 harvester (Innotech, Dottikon, Switzerland). After drying, filters were transferred to scintillation fluid (Betaplate Scint; Wallac U.K., Milton Keynes, U.K.) and counted using a Wallac MicroBeta liquid scintillation counter (1450 MicroBeta; Wallac Oy). Percent apoptosis was quantitated using the following formula: % specific apoptosis = (cpm_spontaneous − cpm_experimental/cpm_spontaneous) × 100. All samples were tested in quadruplicate and are presented in medians with 25–75% interquartiles.

On day 3 after transduction, supernatants from the DC cultures were obtained and stored at −20°C. The soluble murine FasL was determined using a specific ELISA (R & D Systems, Wiesbaden, Germany) according to the manufacturer’s instructions. Results are indicated as means ± SD obtained from at least three independent experiments.

To generate killer APC, iDC and mDC were transduced simultaneously by AdLoxpFasL and AxCANCre. Transduction of DC with either AdLoxpFasL, AxCANCre or AdEGFP was served as the controls. RT-PCR analysis revealed an up-regulated expression of FasL-mRNA in DC simultaneously transduced by AdLoxpFasL and AxCANCre, whereas FasL-mRNA was not detected in AdEGFP- or AxCANCre-transduced DC (Fig. 1 A). As previously observed, FasL-mRNA could be also detected at low levels in AdLoxpFasL singly-transduced DC, suggesting that some spontaneous transgene-expression has occurred using these vectors (34).

To evaluate whether FasL-protein was indeed expressed by AdLoxpFasL and AxCANCre cotransduced DC, FACS-analysis was performed. There was a strong expression of FasL in 80% of mDC 3 days after combined transduction with AdLoxpFasL and AxCANCre. In contrast, FasL could not be detected on the surface of mDC after single transductions by either AdLoxpFasL or AxCANCre. However, <20% of iDC expressed FasL on the surface 72 h after simultaneous transduction with AdLoxpFasL and AxCANCre. These results demonstrated that a combined transduction of these vectors was required for FasL-expression, and that particularly mDC could be efficiently transduced to express FasL (Fig. 1 B).

Expression of Fas on the surface of monocyte-derived DC has been described by several investigators (35, 36, 37, 38). To analyze whether self-destruction might be induced in DC after transduction with FasL, endogenous expression of Fas and FasL was determined in iDC and mDC by FACS. Consistent with previous reports, Fas was expressed in the majority of iDC and mDC, and there was no difference in the expression levels of Fas between both cell populations. In contrast, endogenous human FasL could be detected on the surface of less than 10% of mDC, whereas endogenous FasL was not detectable in iDC (Fig. 2).

These results suggested that Fas-mediated self-destruction might occur in DC as a consequence of FasL expression, limiting the potential usage of these cells. Therefore, apoptosis was detected in iDC and mDC 72 h after adenoviral transduction by annexin V and PI staining. Total numbers of apoptotic cells were determined by calculation of annexin V⁺ and PI⁻ cells (reflecting early apoptosis) together with annexin V⁺ and PI⁺ cells (reflecting late apoptosis/secondary necrosis). There was an apoptosis rate of 91% in iDC 72 h after simultaneous transduction with AdLoxpFasL and AxCANCre, whereas <20% of the singly-transduced iDC were apoptotic. In contrast, ∼70% of FasL-expressing mature DC remained viable at day 3 after transduction with FasL, demonstrating that mDC were protected from Fas-mediated apoptosis despite coexpression of Fas (Fig. 3 A).

To determine the pattern of FasL expression and apoptosis induction, time course experiments were performed at days 1, 2, and 3 after FasL transduction using different concentrations of adenoviral vectors (Fig. 3 B). There was a time- and a dose-dependent expression of FasL in mDC, which peaked 3 days after transduction with AdLoxpFasL and AxCANCre. Expression of FasL or increasing concentrations of adenoviral vectors did not affect apoptosis of mDC, given that the percentage of apoptotic DC was below 10% at all time points tested and virus concentrations used. In contrast, only low levels of FasL were expressed in iDC (<20%) 3 days after transduction using high concentrations of AdLoxpFasL and AxCANCre. In addition, there was a time- and a dose-related induction of apoptosis in iDC. Consequently, the percentage of apoptotic iDC increased above 60% 3 days after FasL transduction using the high virus concentrations.

Next, we sought to analyze the capacity of FasL-expressing DC to induce apoptosis in Fas⁺ target T cells. JAM assays were performed using FasL-expressing DC as effector cells and Jurkat cells, a human T cell lymphoma cell line, as target cells. There was a significant increase in apoptosis in 60% of the Jurkat T cells 5 h after initiation of cocultures with FasL-expressing mDC, whereas there was no increase in apoptosis of Jurkat T cells after coculture with AdEGFP-, AxCANCre-, or AdLoxpFasL-transduced DC lacking murine FasL. The rate of apoptosis induced by FasL-expressing mDC was equivalent to that observed by treatment of Jurkat T cells with an activating anti-Fas Ab. In contrast, iDC exhibited a reduced capacity to induce apoptosis in Jurkat T cells compared with mDC, because apoptosis was detected in <10% of Jurkat T cells cocultured with iDC transduced by AdLoxpFasL and AxCANCre (Fig. 4 A).

In addition, cultures of FasL-expressing mDC and Jurkat T cells were established at different E:T ratios, and cytotoxicity was detected using the JAM assay. There was an apoptosis rate of >80% in Jurkat T cells at the E:T ratio of 1:1 demonstrating the exceptional capacity of mDC to induce apoptosis in Fas⁺ target cells (Fig. 4,B). These results were confirmed by FACS using annexin V and PI staining demonstrating apoptosis in 33% of Jurkat T cells cocultured for 5 h with FasL-expressing DC at E:T 1:5 (Fig. 4 C).

To analyze whether also primary T cells could be eliminated by FasL-expressing DC, additional cytotoxicity experiments were performed. Cocultures of PHA-activated primary T cells and nonactivated primary T cells were established with FasL- or EGFP-transduced mDC at E:T 1:5, and numbers of apoptotic T cells were determined after 5 h by annexin V and PI staining. FACS revealed apoptosis in 47% of the PHA-activated T cells cocultured with FasL-expressing DC, whereas apoptosis could be observed in 14% of PHA-activated T cells cocultured with EGFP-transduced DC. In contrast, apoptosis was not induced in nonactivated T cells cocultured with either FasL- or EGFP-expressing DC. These results indicated that activation of primary T cells was required for induction of apoptosis by FasL-expressing killer DC (Fig. 5).

To analyze whether induction of apoptosis in Fas⁺ target cells by FasL-expressing DC might be affected by sFasL, concentrations of murine sFasL were determined in supernatants of transduced DC using a specific ELISA. There were high levels of sFasL above 3000 pg/ml in culture supernatants of mDC and iDC 72 h after transduction with AdLoxpFasL and AxCANCre, whereas only low concentrations (<25 pg/ml) of sFasL could be detected in supernatants of cultures containing DC after single transduction with these vectors or control cultures containing untreated DC (Fig. 6).

These results suggested that sFasL could be involved in apoptosis induced by FasL-expressing DC; therefore, additional Transwell experiments were performed to test this hypothesis. However, there was no increase in apoptosis in Jurkat T cells cocultured with FasL-expressing mDC when these cell populations were separated by a membrane. In contrast, apoptosis could be induced in 40% of Jurkat T cells by an activating Fas Ab, a positive control that can efficiently induce apoptosis in Jurkat T cells in a soluble form. These results demonstrated that elimination of Fas⁺ target cells by FasL-expressing killer APC required cell-to-cell contact and was not related to production of sFasL (Fig. 7).

The results obtained in this study provide novel information that human monocyte-derived mDC, but not iDC, can be efficiently transduced using adenoviral vectors to express high levels of FasL. These FasL-expressing mDC might act as killer APC to eliminate Fas⁺ target cells including activated primary T cells in a cell-to-cell contact-dependent manner. Thus, the capacity to modify immune responses through FasL-expressing killer APC would provide a significant potential for improvement of therapies for a number of T cell-dependent diseases such as autoimmune diseases, allograft rejection, and chronic infections.

Ag-specific deletion of activated T cells by FasL-expressing APC has been reported previously by us and other investigators. However, all these studies were performed using murine DC, B cells, macrophages, or cell lines as APC (12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 39). Therefore, for future clinical application, it had to be determined whether human DC can be efficiently transduced with FasL and whether these cells would be capable of eliminating Fas⁺ target cells. Phenotypical and functional characterization of DC revealed that DC represent a heterogeneous cell population with diverse functions. These differences have been related to the state of maturation, preparations of DC from different anatomical compartments, and generation of DCs using different species. Because viral gene transfer strategies are required to achieve high transduction rates in human but not murine DC (24), it is particularly important that the differences between murine and human DC populations are also reflected in the capacity to express transduced genes (40, 41).

Adenoviral vectors efficiently transduce human iDC and mDC (24, 42). However, propagation of recombinant adenoviruses encoding FasL has been hampered by induction of self-destruction in host cells leading to low virus titers (25). Therefore, the Cre/Loxp adenovirus vector system was used to enable inducible expression of FasL as previously described (29, 34). Using this adenovirus system, we were able to show that expression of high levels of FasL could be achieved in human mDC in a time- and a virus dose-dependent manner. Highest numbers of FasL-expressing mDC could be detected 3 days after adenoviral transduction and remained high up to day 10 (data not shown). These results are consistent with our previous reports demonstrating that FasL could be induced in murine APC using these adenoviral vectors (22, 28). However, gene transfer strategies to induce tolerance based on viral vectors are limited to in vitro studies as virally transduced DC will present viral Ags on the cell surface leading to activation of virus specific T cells. Thus, if these DC coexpress FasL, virus-specific T cells could be eliminated by Fas-mediated apoptosis leading to suppression of the immune response if the organism is infected with the native virus. Therefore, new methods for high level transduction of human DC are required which are not based on viral vectors.

In contrast to mDC, transduction of iDC with AdLoxpFasL and AxCANCre was less effective. This result was surprising because it has been reported that higher transduction rates could be achieved by adenoviral vectors in iDC compared with mDC (24, 42). However, reduced numbers of iDC were recovered after transduction by AdLoxpFasL and AxCANCre but not control vectors suggesting that iDC were efficiently transduced to express FasL but also were rapidly eliminated by FasL/Fas-mediated apoptosis. This concept was supported by our time course experiments demonstrating that there was a time-dependent increase in apoptotic iDC, which were first annexin V⁺ and PI⁻ and became later annexin V⁺ and PI⁺. The increase in apoptotic iDC was accompanied by an increase in FasL expression in transduced iDC and was also clearly dependent on the dose of AdLoxpFasL and AxCANCre used in these experiments. Together, these results indicated that iDC were rapidly eliminated by Fas-mediated apoptosis shortly after onset of FasL expression, whereas efficiently transduced mDC were protected from apoptosis despite coexpression of Fas and FasL. Therefore, regulatory mechanisms inhibiting the Fas apoptosis signal must be operative downstream of the Fas level in mDC but not iDC. These results are in accordance with previous studies demonstrating that treatment of mDC with activating anti-Fas Abs did not induce apoptosis in these cells despite abundant expression of Fas on the cell surface (37, 43). The resistance of mDC toward Fas-mediated apoptosis has been attributed to the increased expression of anti-apoptotic molecules during DC maturation, and increased levels of c-FLIP and Bcl-x_L have been detected in mDC compared with iDC (35, 38). To prevent Fas-mediated self-destruction, it has been proposed that APC should be transduced with an apoptosis inhibitor in combination with FasL, and protection from Fas-mediated apoptosis could be achieved in murine APC by coexpression of a truncated Fas-associated death domain protein together with FasL using a recombinant vaccinia virus (20). However, our results demonstrate that coexpression of an apoptosis inhibitor is not required if mDC were used for generation of FasL-expressing human killer APC.

It has been reported that human DC are able to directly and effectively mediate apoptotic killing against a wide array of cultured and freshly isolated cancer cells without harming normal cells (44, 45). This antitumor activity has recently been attributed to expression of multiple cytotoxic TNF family ligands including FasL (46). Using adenoviral vectors encoding murine FasL, we were able to discriminate between expressions of endogenous and transduced FasL. FACS revealed small numbers of mDC expressing low levels of endogenous FasL, whereas endogenous FasL could not be detected in iDC. As determined by coculture experiments, untreated iDC or mDC as well as iDC and mDC transduced with control vectors did not exert any detectable cytotoxic activity when Jurkat T cells or activated primary T cells were used as target cells. However, transduction of mDC with AdLoxpFasL and AxCANCre resulted in strong expression of murine FasL, and Fas-mediated apoptosis was efficiently induced in 60% of Jurkat T cells 5 h after initiation of cocultures with transduced FasL-expressing DC as detected by the JAM assay. It is particular important that only a low E:T ratio was required to eliminate the majority of Jurkat T cells, demonstrating the exceptional potency of transduced FasL-expressing DC to act as killer APC, which has been confirmed by staining of target cells with annexin V and PI. Additional cytotoxicity experiments revealed that also primary T cells could be deleted by FasL-expressing DC. However, activation of T cells before establishment of cocultures was required for induction of FasL/Fas-mediated apoptosis, which supported the concept of using killer APC as immunoregulatory cells.

FasL is a type II transmembrane molecule that can be cleaved by specific metalloproteinases to release a 26-kDa soluble form of the protein (47). There are conflicting results regarding the biological role of sFasL. It has been demonstrated that sFasL is able to antagonize the functional activity of membrane FasL (48, 49), whereas other investigators reported that Fas-mediated apoptosis was induced by sFasL in Fas⁺ target cells (50). In contrast to human sFasL, mouse sFasL does not appear to be cytotoxic, although a soluble protein corresponding to the entire extracellular domain of murine FasL has been produced experimentally that does induce apoptosis (51). In addition, the significance of sFasL released by APC transduced with FasL has not been determined. The present results demonstrated that high levels of sFasL could be detected in supernatants obtained from iDC and mDC after transduction with AdLoxpFasL and AxCANCre suggesting that sFasL might be involved in apoptosis. To determine the role of sFasL, additional Transwell experiments were performed. These experiments revealed that apoptosis of Fas⁺ target cells was not mediated by sFasL, but it required cell-to-cell contact between FasL-expressing DC and Fas⁺ target cells. Our results suggest the important role of membrane FasL in this experimental system. However, it has been reported that the capacity of sFasL to induce apoptosis is closely related to the type of cell expressing Fas (52); therefore, cells other than Jurkat T cells might be eliminated by sFasL. In addition, it cannot be excluded from these experiments that membrane FasL was at least partially antagonized by sFasL despite the strong killing activity exhibited by AdLoxpFasL- and AxCANCre-transduced mDC. Moreover, it has been reported that sFasL induced a neutrophil-mediated inflammatory response that might further limit the application of FasL-expressing killer APC (53, 54). Due to these disadvantages, strategies must be developed to prevent the release of sFasL by FasL-transduced DC. This could be achieved by generation of novel gene transfer vectors encoding a FasL gene lacking the metalloproteinase cleavage sites, as previously demonstrated in transfection experiments performed in tumor cell lines (52, 55).

In summary, our results demonstrate that primary human DC can be efficiently transduced by adenoviral vectors to express FasL. Expression of FasL in iDC resulted in self-destruction, whereas FasL-expressing mDC were protected from Fas-mediated apoptosis. These killer APC were able to delete Fas⁺ target cells including activated primary T cells by induction of apoptosis, which required cell-to-cell contact between FasL-expressing DC and target cell, and was not dependent on sFasL. These results support the concept of using human FasL-expressing DC as killer APC for elimination of activated T cells involved in allograft rejection, chronic infections, and autoimmune disease.

We thank Dr. Ulf Müller-Ladner and Dr. Hui-Chen Hsu for careful review of the manuscript.