Dendritic cells (DC) are professional APC endowed with the unique capacity to activate naive T cells. DC also have important effector functions during the innate immune response, such as pathogen recognition and cytokine production. In fact, DC represent the crucial link between innate and adaptive immune responses. However, DC are quite heterogeneous and various subsets endowed with specific pathogen recognition mechanisms, locations, phenotypes, and functions have been described both in rodents and in humans. A series of studies indicated that rodent as well as human DC could also mediate another important innate function, i.e., cell-mediated cytotoxicity, mostly toward tumor cells. In this article, we will review the phenotypes of these so-called killer DC, their killing mechanism, and putative implication in the immune response.

Cell-mediated cytotoxicity is an important effector function of the immune system and plays a crucial role in the elimination of virus-infected and tumor cells. Besides NK and CD8+ T cells, which are highly specialized killer cells during the innate and adaptive responses, respectively, cells of the myeloid lineage also exhibit cytotoxic potential to various target cells using mechanisms that might also be used to eliminate pathogens. For instance, the killing activity of activated macrophages toward tumor cells has been described for many years, and mechanisms of killing involve TNF-α, NO, or reactive oxygen species (1). Dendritic cells (DC)4 are well known as professional APC able to activate naive T cells and thus are critical for the development of adaptive immune responses (2). DC also have important effector functions during the innate immune response. In fact, multiple subsets of DC have been described that are endowed with specific pathogen recognition mechanisms, locations, phenotypes, and types of Ag presentation and cytokine production (3). An in vitro killing activity has been described in human and rodent conventional DC (cDC) and/or plasmacytoid DC (pDC) generated in vitro or in vivo (Table I). Intriguingly, the molecular mechanisms of killing used by these killer DC (KDC) also appear quite diverse (Table I). Apart from its putative direct antitumor activity, this unusual DC function could facilitate cellular Ag uptake and therefore Ag presentation.

Table I.

DCs with killing properties: subsets and mechanisms

SpeciesSubset/PhenotypeMechanism of KillingInducersTarget CellsRef.
Human CD34-derived DC TRAIL IFN-β HL60 and Reh 39  
 MoDC TNF-α LPS; IFN-γ Breast tumor cells 40  
   CD40L MDA231 38  
  TRAIL Measles virus MDA231 38  
   CMV Jurkat, autologous PBMC 63  
   LPS; IFN-γ Hematological tumor cells 34  
   IFN-β HL60, Reh 39  
   dsRNA MDA231 38  
   Type I IFN + GM-CSF Jurkat 37  
  None HPV-transformed keratinocytes 34  
   None Various tumor cells 31  
  FasL CMV Jurkat; autologous PBMC 63  
  Simultaneous engagement of various death HIV MDA231, T cell line 42  
   ligands None Various tumor cells, Jurkat, endothelial cells 2944  
 Blood CD11c+ mDC TRAIL IFN-α or γ Various tumor cells 36  
  Perforin-Gzm Imiquimod K562 49  
 Blood DC (Lin-CD4+ HLA-DR+TNF-α, LT-α1β2, FasL, TRAIL None Various tumor cells, Jurkat, endothelial cells 3144  
 Blood M-DC8+ mDC TNF-α IFN-γ Various tumor cell lines 47  
 Blood pDC GrzB IL-3 and CD40L (CpG) K562 a 
  TRAIL Imiquimod Jurkat 49 
   Influenza virus; CpG; R848 A549, Mel 43  
   HIV, Imiquimod IFN1 T cells 48  
 Thymic DC FasL and TNF-α HIV Activated PBMC, T cells 62  
Mouse BMDC FasL None Jurkat, alloactivated T cells 4  
  FasL and TRAIL (partially) Spontaneously mature A20, EL-4 5  
  TRAIL (partially) LPS T cell 6  
 LNDC FasL Influenza virus T cell clone 8  
 Bitypic NK/DCb LCMV+ aCD40L RMAS and YAC 13  
 NKDCb Perforin-Gzm None YAC, CHO 14  
Rat BMDC NO LPS; IFN-γ Various tumor cells 29  
   NKG2D Rat histiocytic tumor 27  
 Splenic CD103+ DC Perforin-Gzm Spontaneously mature YAC 22  
  TNF-α NKG2D Rat histiocytic tumor 26  
 Spleen and LN CD103+CD4+ DC None Various tumor cells, endothelial cells 2324  
 Thymic DC NO None Thymocytes 28  
SpeciesSubset/PhenotypeMechanism of KillingInducersTarget CellsRef.
Human CD34-derived DC TRAIL IFN-β HL60 and Reh 39  
 MoDC TNF-α LPS; IFN-γ Breast tumor cells 40  
   CD40L MDA231 38  
  TRAIL Measles virus MDA231 38  
   CMV Jurkat, autologous PBMC 63  
   LPS; IFN-γ Hematological tumor cells 34  
   IFN-β HL60, Reh 39  
   dsRNA MDA231 38  
   Type I IFN + GM-CSF Jurkat 37  
  None HPV-transformed keratinocytes 34  
   None Various tumor cells 31  
  FasL CMV Jurkat; autologous PBMC 63  
  Simultaneous engagement of various death HIV MDA231, T cell line 42  
   ligands None Various tumor cells, Jurkat, endothelial cells 2944  
 Blood CD11c+ mDC TRAIL IFN-α or γ Various tumor cells 36  
  Perforin-Gzm Imiquimod K562 49  
 Blood DC (Lin-CD4+ HLA-DR+TNF-α, LT-α1β2, FasL, TRAIL None Various tumor cells, Jurkat, endothelial cells 3144  
 Blood M-DC8+ mDC TNF-α IFN-γ Various tumor cell lines 47  
 Blood pDC GrzB IL-3 and CD40L (CpG) K562 a 
  TRAIL Imiquimod Jurkat 49 
   Influenza virus; CpG; R848 A549, Mel 43  
   HIV, Imiquimod IFN1 T cells 48  
 Thymic DC FasL and TNF-α HIV Activated PBMC, T cells 62  
Mouse BMDC FasL None Jurkat, alloactivated T cells 4  
  FasL and TRAIL (partially) Spontaneously mature A20, EL-4 5  
  TRAIL (partially) LPS T cell 6  
 LNDC FasL Influenza virus T cell clone 8  
 Bitypic NK/DCb LCMV+ aCD40L RMAS and YAC 13  
 NKDCb Perforin-Gzm None YAC, CHO 14  
Rat BMDC NO LPS; IFN-γ Various tumor cells 29  
   NKG2D Rat histiocytic tumor 27  
 Splenic CD103+ DC Perforin-Gzm Spontaneously mature YAC 22  
  TNF-α NKG2D Rat histiocytic tumor 26  
 Spleen and LN CD103+CD4+ DC None Various tumor cells, endothelial cells 2324  
 Thymic DC NO None Thymocytes 28  
a

K. Palucka and J. Banchereau, personal communication.

b

Similar to IKDC, bitypic NK/DC and NKDC are in fact a a subset of NK cells (17–19).

Several groups reported that in vitro generated myeloid DC (mDC) as well as various subsets of in vivo generated cDC subsets can induce apoptosis in T cells and tumor cells in vitro. Bone marrow-derived DC have been reported to kill T or tumor cells using Fas ligand (FasL) (4, 5) or TRAIL (6). Langerhans cells (7) and lymph node (LN) CD11c+ cells from influenza-infected mice can express functional FasL (8), yet the precise phenotype of FasL+ DC was not reported in the latter study. Splenic resident CD8α+ cDC were reported to induce Fas-mediated apoptosis in CD4+ T cells (9); however, specific FasL expression in this DC subset was not further reported. The same subset constitutively expresses IDO and can induce IDO-dependent apoptosis of T cells (10). A pDC subset also expresses IDO and plays a role in tumor tolerance, but through regulatory T cells rather than the induction of T cell apoptosis (11). Some inflammatory DC that play an important role in the response to Listeria express two potentially cytotoxic molecules, TNF-α and inducible NO synthase (TIP-DC) (12).

More recently, specific subsets of cells exhibiting both NK and DC features and called NK/DC (13), NKDC (14), and IFN-producing KDC (IKDC) (15, 16) were described and claimed to exhibit four major properties: 1) expression of both DC (CD11c) and NK (NK1.1, CD49b) markers; 2) TRAIL and/or perforin-granzyme (Gzm)-mediated activity; 3) strong production of IFN-γ upon stimulation by tumor, cytokines, and CpG; and 4) APC function following appropriate stimulation. Among these subsets, IKDC were the most phenotypically well defined. IKDC express B220 but not the pDC-specific surface molecule PDCA1. IKDC were shown to mediate the potent TRAIL-dependent, antitumoral effect observed upon treatment by imatinib mesylate plus IL-2 in a mouse melanoma model (16). The NK-related functions of these hybrid cells were more potent than those of NK cells, whereas their APC function was actually lower than that of DC (15) or simply not demonstrated (16). The cytolytic activity of IKDC was in fact transient after TLR9 stimulation and down-regulated after prolonged stimulation (15). Finally, the fact that IKDC were present in Rag2−/− IL2Rγ−/− mice suggested they were developmentally unrelated to NK cells (16). Recent data have challenged the notion that IKDC belong to the DC family and indicate they are more likely a subset of B220+ and probably activated NK cells (17, 18, 19). In fact, IKDC have a CD3NKp46+ phenotype (17), which was put forward as the definition of NK cells (20). More importantly, these studies reported that the so-called IKDC, even the small proportion of these cells that express low amounts of MHC class II (MHCII), were very poor APC even after stimulation (18). In addition, the fact that CD49b is apparently expressed by a small number of CD11chigh cDC (18) might lead to cDC contamination when both these markers are used for cell sorting (15). In contrast with the initial study, putative IKDC and NK cells, but not cDC or pDC, were virtually absent in Rag2−/−IL-2Rγ−/− mice (17, 18). Moreover, the reported ability of IKDC to produce type I IFN (15) was not confirmed and likely due to pDC contamination (21). Finally, the capacity of NKDC (defined as the NK1.1+ subset of CD11c+ cells), but not NK cells, to produce IFN-γ in response to TLR9 plus IL-4 (14), might be due to pDC or cDC contamination, which could activate NK through TLR9-induced IL-12 production or to differential TLR9 expression between NK subsets.

The first evidence that DC can kill tumor cells was obtained in the rat. Following spontaneous maturation in vitro, splenic CD103+ DC were shown to exhibit a Ca2+-dependent cytotoxicity against select target cells in vitro (22). We later identified a subset of immature CD11b/c+ DC with spontaneous killing properties in lymphoid organs (23, 24). Several features of these KDC argue against the possibility that they represent of subset of NK cells: 1) KDC exhibit a homogeneous MHCII+CD11b+CD103highCD4NKp46 phenotype (these DC are also referred to as CD4 DC), whereas rat NK cells are MHCIICD103NKp46+ (our unpublished observations); 2) KDC induce rapid caspase-independent apoptosis-like cell death in various tumor cells and endothelial cells using a mechanism independent of perforin-Gzm, FasL, TRAIL, or TNF-α (23); 3) unlike NK cells, KDC produce very low if any IFN-γ (25); 4) target cell killing is immediately followed by their phagocytosis by KDC (24); and 5) upon TLR or CD40L stimulation, KDC produce large amounts of IL-12, up-regulate MHCII, CD80, and CD86 molecules, and become potent APC in vitro (25) while strongly down-regulating their tumoricidal and phagocytic properties (24). In vivo, tumor regression was observed following the vaccination of osteosarcoma-bearing rats with KDC that were previously allowed to kill and phagocytose a tumor-derived cell line in vitro, suggesting that KDC can efficiently present Ag from their victims (our unpublished observations). CD103+ DC with TNF-α-mediated cytotoxicity have also been isolated by others from the spleen and adjacent LN of tumor-bearing rats (26). Tumor cell recognition appeared to be mediated by NKG2D on DC (26), which triggered maturation and production of cytokines and NO that participate in the killing of tumor cells in vitro (27). Thymic DC (28) and LPS- or IFN-γ-activated bone marrow-derived DC also exhibited NO-dependent killing properties after prolonged contact with targets (29).

Cytotoxic activity of in vitro generated myeloid human DC has been reported by many independent groups, yet various mechanisms and inducing signals were described for the same cells, probably because of variability in culture conditions. Immature DC generated in vitro from blood monocytes (MoDC) were reported to induce apoptosis in hematopoietic (30, 31, 32) and nonhematopoietic tumor cells (30, 32, 33) without damaging normal cells (34, 35). A large number of studies further showed that activation/maturation of MoDC with type I IFN (36, 37, 38, 39), IFN-γ (36), LPS (40), dsRNA, or various viruses (38, 41, 42, 43), but not CD40L (30, 38, 44), enhanced or was required to induce cytotoxicity. This activity was mostly mediated by TRAIL, which could in fact be induced by type I IFN produced in an autocrine fashion by DC. In addition, MoDC generated in vitro by GM-CSF and IFN-α instead of IL-4 exhibited expression of TRAIL and GzmB and killed K562 cells (45). DC generated from CD34 cells also acquired killing potential upon IFN-β stimulation (39). Cooperation between TRAIL, FasL, TNF-α, and lymphotoxin α1β2 in the killing mechanisms of MoDC has also been demonstrated (42, 44). Additional and to date unknown mechanisms of MoDC-mediated killing have also been reported (31, 32, 34).

Fewer studies reported killing by in vivo generated DC. Activated Langerhans cells were reported to express membrane FasL, yet their killing function was not evaluated (46). Two major subsets of DC are found in peripheral blood: 1) CD11c+CD123 or mDC, which can be further divided into three populations based on their expression of CD1c, CD141, and CD16; and 2) CD11cCD123highCD303+ cells that are pDC. Similar to MoDC and monocytes, IFN-γ- or IFN-α-stimulated mDC can express TRAIL and become killers in vitro (36). Unstimulated blood LinCD4+HLA-DR+ DC were later reported to induce apoptosis in many cancer cells but not normal cells, with the exception of proliferating endothelial cells (30, 44). The M-DC8 mAb defines an inflammatory subset of CD16+ mDC endowed with a TNF-α-dependent cytotoxic activity against specific epithelial tumor cells (47). Whether purified CD1c+ and CD141+ mDC also exhibit killing properties is unknown. pDC were also shown to express functional TRAIL upon virus, TLR7/8, or TLR9 stimulation (43, 48). A recent report showed that, following local treatment of skin carcinoma with a TLR7/8 agonist (imiquimod), TRAIL-expressing pDC infiltrated the tumor, whereas peritumoral mDC expressed perforin and GzmB (49). In vitro, TLR7–8 agonists induced 10–15% of blood mDC to express perforin plus GzmB and to kill K562 cells. GzmB was in fact constitutive in 70% of pDC but not mDC and was down-regulated upon TLR7 triggering, which simultaneously induced functional TRAIL on pDC (49). This is consistent with a previous report describing the expression of GzmB in pDC (50); yet, the role of GzmB in the absence of perforin in pDC remains to be elucidated. Interestingly, in addition to its proapoptotic properties, GzmB could also play a role in extracellular remodeling (51). Whether GzmB in pDC is expressed in lysosomes and whether GzmB and TRAIL are expressed by the same pDC in a time- and activation-dependent manner is unknown. In fact, only the CD2+ subset of pDC could secrete GzmB and kill K562 cells when activated with IL-3 plus CD40 ligand (K. Palucka and J. Banchereau, unpublished observations). Killer pDC have not been described in mice or rats, although we found that rat pDC strongly expressed GzmB upon TLR9 activation (our unpublished observations). Finally, a subset of human DC specifically expressing functional IDO has also been described, although the phenotype of these cells has not been fully characterized (52).

As described previously, some DC might exhibit innate cytotoxic properties while other populations might become killer only after appropriate stimulation. Moreover, a few studies suggest that DC can kill normal T cells. Obviously, these various situations might lead to very different fates of the immune response (Fig. 1).

FIGURE 1.

Hypotheses concerning the role of KDC. DC with killing properties have been described in human and rodents. This cytotoxicity can be natural or induced on various DC subsets, including pDC, by inducers that are mainly related to virus and antiviral response. KDC can play a role in killing tumor or infected cells by using TRAIL, FasL, perforin-Gzm, or NO. Mechanisms of target cell recognition are mostly unknown. After killing, DC could phagocytose apoptotic cells for cross-presenting Ags to CD8+ T cells, provided they receive a strong maturation signal. By killing T cells in an Ag-dependent or independent manner using FasL, TRAIL, or IDO, KDC might, in contrast, participate in T cell tolerance. This function could be exploited by viruses and tumors to escape immune response. Finally, some immature KDC could be specialized in the killing of senescent cells in tissues, a function that could allow these DC to efficiently phagocytose self-Ag for the purpose of inducing/maintaining self-tolerance.

FIGURE 1.

Hypotheses concerning the role of KDC. DC with killing properties have been described in human and rodents. This cytotoxicity can be natural or induced on various DC subsets, including pDC, by inducers that are mainly related to virus and antiviral response. KDC can play a role in killing tumor or infected cells by using TRAIL, FasL, perforin-Gzm, or NO. Mechanisms of target cell recognition are mostly unknown. After killing, DC could phagocytose apoptotic cells for cross-presenting Ags to CD8+ T cells, provided they receive a strong maturation signal. By killing T cells in an Ag-dependent or independent manner using FasL, TRAIL, or IDO, KDC might, in contrast, participate in T cell tolerance. This function could be exploited by viruses and tumors to escape immune response. Finally, some immature KDC could be specialized in the killing of senescent cells in tissues, a function that could allow these DC to efficiently phagocytose self-Ag for the purpose of inducing/maintaining self-tolerance.

Close modal

DC with killing properties appear to be mostly specialized in the killing of tumor cells, suggesting that these cells can likely recognize tumor cells using specific receptors. NKG2D has been identified in rat KDC (26) but is not expressed on resting or activated human mDC or pDC (49). NKp44 was found on some human tonsil and blood pDC (53), although NKp44 cross-linking failed to induce lytic activity. To our knowledge, the expression of other cytotoxicity-triggering receptors has not been reported for human DC (30). However, receptor-mediated tumor recognition is probably required to trigger exocytosis of cytotoxic granules in DC that use this pathway (22, 49). Interestingly, natural cytotoxicity receptors expressed by NK cells (NKp30, NKp46, NKG2-D) appeared not only to mediate target cell recognition but also to allow Ag internalization into MHCII peptide loading compartments (54).

What could the role of KDC be in immune responses to tumors? First, KDC could directly participate in tumor elimination. However, DC are rare cells and would need to be recruited in large numbers into tumors to play a direct role in regression. A positive correlation was observed between tumor-infiltrating DC aggregation and apoptosis in lung cancers (55). However, the levels of tumor infiltration did not always positively correlate with outcome and depended on the type, maturation status, and location of infiltrating DC (56). A recent study shed light on the potential direct cytotoxic activity of DC in human cancer (49). As described above, upon topical treatment with imiquimod, skin basal cell carcinoma regressed and exhibited enhanced infiltration by inflammatory cells, including mDC and pDC. A number of mDC were shown to express perforin-GzmB or TRAIL, whereas pDC expressed TRAIL. Yet, both cell types were 5-fold less abundant than CTL, and most of the TRAIL-expressing cells were in fact T cells. More surprising was the finding that the majority of perforin-GzmB-expressing cells were mDC. In addition, these DC coexpressed TNF-α and inducible NO synthases, like TIP-DC, which makes them exceptionally well-equipped putative KDC. However, their role in killing tumor cells in vivo was not demonstrated. Whether the induction of TRAIL on DC in this setting is dependent on TLR7-induced type I IFN production by pDC (48) needs to be addressed. In contrast, IDO-expressing DC in tumors or in draining LN could play an important role in tumor escape by inducing apoptosis of tumor-specific T cells (57). It is interesting to note that a TLR7/8 agonist appeared to impair IDO expression in MoDC (58).

Second, the killing activity of DC could allow them to rapidly phagocytose material from apoptotic cells and to present and/or cross-present Ag to T cells. Both human mDC (59) and pDC (60) efficiently phagocytose apoptotic cell fragments for cross-presentation to CD8+ T cells. The tumoricidal properties of DC might favor the phagocytosis of apoptotic tumor cells by DC rather than by phagocytes. In the rat, we found that KDC started to phagocytose apoptotic cell fragments in the minutes following induction of cell death, suggesting that both events are intimately linked (24). However, the induction of cytotoxicity in human DC by TLR or IFN should trigger maturation and therefore decrease their phagocytic activity. For understanding the nature of immune response induced by these DC (tolerogenic vs immunogenic), it will be critical to determine whether tumor cell recognition by killer DC triggers DC maturation in vivo. The fact that some KDC could induce lysis of proliferating endothelial cells also suggests a potential role in limiting tumor blood vessel formation (30).

Virus-related signals such as TLR7 and 9 or type I IFN appear to play an important role in the induction of cytotoxic properties in DC that could have different effects during viral infection. Measles, influenza, and HIV infection were shown to induce MoDC and/or pDC to express TRAIL and to efficiently kill target cells (38, 43, 48). In addition, human MoDC were reported to kill papillomavirus-transformed keratinocytes through an unknown recognition mechanism (34). The lytic properties of DC could in this case represent a way to eliminate infected cells that could subsequently be phagocytosed by DC and cross-presented to CD8+ T cells. In contrast, DC cytotoxicity could limit T cell expansion and immunopathology or might be used by pathogens to escape immune responses. Along these lines, LN DC purified from mice infected with a high dose of influenza induced FasL-mediated apoptosis of proliferating CD8+ T cells (8). Similarly, HIV-1-infected thymic DC released uncharacterized cytotoxic agents that induced lysis of CD4+ and CD8+ T cells (61, 62). CMV- and HIV-infected human MoDC (42, 63) and HIV-infected pDC (48) also triggered apoptosis of infected and uninfected activated CD4+ T cells via TRAIL and FasL. These data suggest that viruses might exploit the lytic properties of DC to deplete T cells in the early stage of infection.

As described previously, by killing T cells, DC could participate in clonal deletion or could limit T cell expansion. Thymic DC were reported to produce NO upon contact with self-antigens and alloantigens and to induce apoptosis of thymocytes (28). Some DC might also use their cytolytic function to acquire autoantigens in peripheral tissues. Immature DC are thought to uptake apoptotic cells that appear during tissue renewal, and to migrate to draining LN where they present autoantigens for the induction of self-tolerance (64). However, the importance of this mechanism could be limited by the fact that apoptotic cells can be phagocytosed by other cells and that some tissues have low rates of turnover. Huang et al. observed that only a subset of migratory rat DC in afferent lymph transported apoptotic epithelial fragments to LN (65). Intriguingly, the phenotype of these DC is similar to that of the KDC described in rat spleen and LN (23). In fact, such CD4 DC were found in all lymphoid organs and most nonlymphoid tissues (our unpublished observations), although tissue CD4 DC cytotoxicity has not been assessed. This raises the hypothesis that CD4 DC could use their killing activity to phagocytose tissue cells in the steady state. The fact that CD4 DC killed endothelial cells in vitro suggest a capacity to lyse “normal” cells (24). It is possible that CD4 DC recognize “kill me” signals in senescent cells, which would thereby be “euthanized” and immediately engulfed by the same DC for transport to draining LN. This hypothesis implies that target cell killing and phagocytosis would not induce full DC maturation. Such a function could finally be exploited by a tumor for promoting tolerance and therefore escape immune response.

Despite being described 10 years ago, the cytotoxic properties of DC have received little attention. Human and rodent studies clearly indicate that many different types of DC, including pDC, can acquire cytolytic potential. The fact that DC can express Gzm and perforin in DC should not be a surprise, as these conserved molecules were shown to be expressed by nonlymphoid cells in different ancestral species as well as in rat macrophages, for instance (66). The induction of TRAIL expression by IFN is also a general phenomenon described for T cells, monocytes, mDC, and pDC. It is also possible that some specific subsets of DC have a constitutive killing activity, although the presence of NK cells in these populations should be carefully examined, keeping in mind that certain human NK cells have been shown to exhibit APC-like functions (54, 67). Indeed, human NK cells express substantial levels of MHCII and various costimulatory molecules and, more surprisingly, were shown to display efficient Ag capture and presentation capacities in vitro (54).

The role of the cytotoxic activity of DC in immune responses is still unknown, and a demonstration that DC can directly kill target cells in vivo is still lacking. More work is also needed to understand the molecular pathways leading to the acquisition and regulation of killing properties and the mechanisms of tumor recognition by DC. We hypothesize that some DC used their killing potential for the purpose of acquiring cellular Ags. This might be important for linking innate and adaptive responses in situations in which NK cells are unable to kill tumors or infected cells, for instance. However, an important question will be to determine whether target cell recognition and/or killing induces DC maturation. If not, these DC might in fact promote tolerance rather than immunity. The induction of killing properties in either mDC or pDC could be harnessed for therapeutic use by local or systemic administration of TLR7/8 or TLR9 ligands or by intratumoral injection of in vitro generated cytotoxic DC (68). Alternatively, this additional function of DC could provide an efficient way to load tumor Ag into DC for vaccine strategies. The relevance of this unusual feature of DC should be further explored in light of these potential therapeutic applications.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

Our work was supported by Institut National de la Santé et de la Recherche Médicale and the Association pour la Recherche sur le Cancer.

4

Abbreviations used in this paper: DC, dendritic cell; cDC, conventional DC; FasL, Fas ligand; Gzm, granzyme; IKDC, IFN-producing killer DC; KDC, killer DC; LN, lymph node; MHCII, MHC class II; MoDC, monocyte-derived DC; mDC, myeloid DC; pDC, plasmacytoid DC.

1
Aliprantis, A. O., G. Diez-Roux, L. C. Mulder, A. Zychlinsky, R. A. Lang.
1996
. Do macrophages kill through apoptosis?.
Immunol. Today
17
:
573
-576.
2
Banchereau, J., F. Briere, C. Caux, J. Davoust, S. Lebecque, Y. J. Liu, B. Pulendran, K. Palucka.
2000
. Immunobiology of dendritic cells.
Annu. Rev. Immunol.
18
:
767
-811.
3
Shortman, K., Y. J. Liu.
2002
. Mouse and human dendritic cell subtypes.
Nat. Rev. Immunol.
2
:
151
-161.
4
Lu, L., S. Qian, P. A. Hershberger, W. A. Rudert, D. H. Lynch, A. W. Thomson.
1997
. Fas ligand (CD95L) and B7 expression on dendritic cells provide counter-regulatory signals for T cell survival and proliferation.
J. Immunol.
158
:
5676
-5684.
5
Huang, J., T. Tatsumi, E. Pizzoferrato, N. Vujanovic, W. J. Storkus.
2005
. Nitric oxide sensitizes tumor cells to dendritic cell-mediated apoptosis, uptake, and cross-presentation.
Cancer Res.
65
:
8461
-8470.
6
Yu, Y., S. Liu, W. Wang, W. Song, M. Zhang, W. Zhang, Z. Qin, X. Cao.
2002
. Involvement of tumour necrosis factor-α-related apoptosis-inducing ligand in enhanced cytotoxicity of lipopolysaccharide-stimulated dendritic cells to activated T cells.
Immunology.
106
:
308
-315.
7
Shibaki, A., S. I. Katz.
2001
. Activation through CD40 ligation induces functional Fas ligand expression by Langerhans cells.
Eur. J. Immunol.
31
:
3006
-3015.
8
Legge, K. L., T. J. Braciale.
2005
. Lymph node dendritic cells control CD8+ T cell responses through regulated FasL expression.
Immunity.
23
:
649
-659.
9
Suss, G., K. Shortman.
1996
. A subclass of dendritic cells kills CD4 T cells via Fas/Fas-ligand-induced apoptosis.
J. Exp. Med.
183
:
1789
-1796.
10
Fallarino, F., C. Vacca, C. Orabona, M. L. Belladonna, R. Bianchi, B. Marshall, D. B. Keskin, A. L. Mellor, M. C. Fioretti, U. Grohmann, P. Puccetti.
2002
. Functional expression of indoleamine 2,3-dioxygenase by murine CD8α+ dendritic cells.
Int. Immunol.
14
:
65
-68.
11
Sharma, M. D., B. Baban, P. Chandler, D. Y. Hou, N. Singh, H. Yagita, M. Azuma, B. R. Blazar, A. L. Mellor, D. H. Munn.
2007
. Plasmacytoid dendritic cells from mouse tumor-draining lymph nodes directly activate mature Tregs via indoleamine 2,3-dioxygenase.
J. Clin. Invest.
117
:
2570
-2582.
12
Serbina, N. V., T. P. Salazar-Mather, C. A. Biron, W. A. Kuziel, E. G. Pamer.
2003
. TNF/iNOS-producing dendritic cells mediate innate immune defense against bacterial infection.
Immunity.
19
:
59
-70.
13
Homann, D., A. Jahreis, T. Wolfe, A. Hughes, B. Coon, M. J. van Stipdonk, K. R. Prilliman, S. P. Schoenberger, M. G. von Herrath.
2002
. CD40L blockade prevents autoimmune diabetes by induction of bitypic NK/DC regulatory cells.
Immunity.
16
:
403
-415.
14
Pillarisetty, V. G., S. C. Katz, J. I. Bleier, A. B. Shah, R. P. Dematteo.
2005
. Natural killer dendritic cells have both antigen presenting and lytic function and in response to CpG produce IFN-γ via autocrine IL-12.
J. Immunol.
174
:
2612
-2618.
15
Chan, C. W., E. Crafton, H. N. Fan, J. Flook, K. Yoshimura, M. Skarica, D. Brockstedt, T. W. Dubensky, M. F. Stins, L. L. Lanier, et al
2006
. Interferon-producing killer dendritic cells provide a link between innate and adaptive immunity.
Nat. Med.
12
:
207
-213.
16
Taieb, J., N. Chaput, C. Menard, L. Apetoh, E. Ullrich, M. Bonmort, M. Pequignot, N. Casares, M. Terme, C. Flament, et al
2006
. A novel dendritic cell subset involved in tumor immunosurveillance.
Nat. Med.
12
:
214
-219.
17
Vosshenrich, C. A., S. Lesjean-Pottier, M. Hasan, O. R. Goff, E. Corcuff, O. Mandelboim, J. P. Di Santo.
2007
. CD11cloB220+ interferon-producing killer dendritic cells are activated natural killer cells.
J. Exp. Med.
204
:
2569
-2578.
18
Caminschi, I., F. Ahmet, K. Heger, J. Brady, S. L. Nutt, D. Vremec, S. Pietersz, M. H. Lahoud, L. Schofield, D. S. Hansen, et al
2007
. Putative IKDCs are functionally and developmentally similar to natural killer cells, but not to dendritic cells.
J. Exp. Med.
204
:
2569
-2578.
19
Blasius, A. L., W. Barchet, M. Cella, M. Colonna.
2007
. Development and function of murine B220+CD11c+NK1.1+ cells identify them as a subset of NK cells.
J. Exp. Med.
204
:
2561
-2578.
20
Walzer, T., S. Jaeger, J. Chaix, E. Vivier.
2007
. Natural killer cells: from CD3NKp46+ to post-genomics meta-analyses.
Curr. Opin. Immunol.
19
:
365
-372.
21
Vremec, D., M. O'Keeffe, H. Hochrein, M. Fuchsberger, I. Caminschi, M. Lahoud, K. Shortman.
2007
. Production of interferons by dendritic cells, plasmacytoid cells, natural killer cells, and interferon-producing killer dendritic cells.
Blood
109
:
1165
-1173.
22
Josien, R., M. Heslan, J. P. Soulillou, M. C. Cuturi.
1997
. Rat spleen dendritic cells express natural killer cell receptor protein 1 (NKR-P1) and have cytotoxic activity to select targets via a Ca2+-dependent mechanism.
J. Exp. Med.
186
:
467
-472.
23
Trinite, B., C. Voisine, H. Yagita, R. Josien.
2000
. A subset of cytolytic dendritic cells in rat.
J. Immunol.
165
:
4202
-4208.
24
Trinite, B., C. Chauvin, H. Peche, C. Voisine, M. Heslan, R. Josien.
2005
. Immature CD4CD103+ rat dendritic cells induce rapid caspase-independent apoptosis-like cell death in various tumor and nontumor cells and phagocytose their victims.
J. Immunol.
175
:
2408
-2417.
25
Hubert, F.-X., C. Voisine, C. Louvet, A. Ouabed, M. Heslan, R. Josien.
2006
. Differential pattern recognition receptor expression but stereotyped responsiveness in rat spleen dendritic cell subsets.
J. Immunol.
177
:
1007
-1016.
26
Alli, R., B. Savithri, S. Das, C. Varalakshmi, N. Rangaraj, A. Khar.
2004
. Involvement of NKR-P2/NKG2D in DC-mediated killing of tumor targets: indicative of a common, innate, target-recognition paradigm?.
Eur. J. Immunol.
34
:
1119
-1126.
27
Srivastava, R. M., C. Varalakshmi, A. Khar.
2007
. Cross-linking a mAb to NKR-P2/NKG2D on dendritic cells induces their activation and maturation leading to enhanced anti-tumor immune response.
Int. Immunol.
19
:
591
-607.
28
Aiello, S., M. Noris, G. Piccinini, S. Tomasoni, F. Casiraghi, S. Bonazzola, M. Mister, M. H. Sayegh, G. Remuzzi.
2000
. Thymic dendritic cells express inducible nitric oxide synthase and generate nitric oxide in response to self- and alloantigens.
J. Immunol.
164
:
4649
-4658.
29
Nicolas, A., D. Cathelin, N. Larmonier, J. Fraszczak, P. E. Puig, A. Bouchot, A. Bateman, E. Solary, B. Bonnotte.
2007
. Dendritic cells trigger tumor cell death by a nitric oxide-dependent mechanism.
J. Immunol.
179
:
812
-818.
30
Janjic, B. M., G. Lu, A. Pimenov, T. L. Whiteside, W. J. Storkus, N. L. Vujanovic.
2002
. Innate direct anticancer effector function of human immature dendritic cells. I. Involvement of an apoptosis-inducing pathway.
J. Immunol.
168
:
1823
-1830.
31
Ayres, F. M., M. Narita, M. Takahashi, T. Yano, A. Liu, K. Toba, T. Furukawa, Y. Aizawa.
2004
. Human dendritic cells mediate anti-tumor activity against hematopoietic tumor cells without direct contact and Fas/FasL killing pathway.
Oncol. Rep.
11
:
1017
-1023.
32
Vanderheyde, N., E. Aksoy, Z. Amraoui, P. Vandenabeele, M. Goldman, F. Willems.
2001
. Tumoricidal activity of monocyte-derived dendritic cells: evidence for a caspase-8-dependent, Fas-associated death domain-independent mechanism.
J. Immunol.
167
:
3565
-3569.
33
Joo, H. G., T. P. Fleming, Y. Tanaka, T. J. Dunn, D. C. Linehan, P. S. Goedegebuure, T. J. Eberlein.
2002
. Human dendritic cells induce tumor-specific apoptosis by soluble factors.
Int. J. Cancer
102
:
20
-28.
34
Hubert, P., S. L. Giannini, A. Vanderplasschen, E. Franzen-Detrooz, N. Jacobs, J. Boniver, P. Delvenne.
2001
. Dendritic cells induce the death of human papillomavirus-transformed keratinocytes.
FASEB J.
15
:
2521
-2523.
35
Shi, J., K. Ikeda, N. Fujii, E. Kondo, K. Shinagawa, F. Ishimaru, K. Kaneda, M. Tanimoto, X. Li, Q. Pu.
2005
. Activated human umbilical cord blood dendritic cells kill tumor cells without damaging normal hematological progenitor cells.
Cancer Sci.
96
:
127
-133.
36
Fanger, N. A., C. R. Maliszewski, K. Schooley, T. S. Griffith.
1999
. Human dendritic cells mediate cellular apoptosis via tumor necrosis factor-related apoptosis-inducing ligand (TRAIL).
J. Exp. Med.
190
:
1155
-1164.
37
Santini, S. M., C. Lapenta, M. Logozzi, S. Parlato, M. Spada, T. Di Pucchio, F. Belardelli.
2000
. Type I interferon as a powerful adjuvant for monocyte-derived dendritic cell development and activity in vitro and in Hu-PBL-SCID mice.
J. Exp. Med.
191
:
1777
-1788.
38
Vidalain, P. O., O. Azocar, H. Yagita, C. Rabourdin-Combe, C. Servet-Delprat.
2001
. Cytotoxic activity of human dendritic cells is differentially regulated by double-stranded RNA and CD40 ligand.
J. Immunol.
167
:
3765
-3772.
39
Liu, S., Y. Yu, M. Zhang, W. Wang, X. Cao.
2001
. The involvement of TNF-α-related apoptosis-inducing ligand in the enhanced cytotoxicity of IFN-β-stimulated human dendritic cells to tumor cells.
J. Immunol.
166
:
5407
-5415.
40
Manna, P. P., T. Mohanakumar.
2002
. Human dendritic cell mediated cytotoxicity against breast carcinoma cells in vitro.
J. Leukocyte Biol.
72
:
312
-320.
41
Vidalain, P. O., O. Azocar, B. Lamouille, A. Astier, C. Rabourdin-Combe, C. Servet-Delprat.
2000
. Measles virus induces functional TRAIL production by human dendritic cells.
J. Virol.
74
:
556
-559.
42
Lichtner, M., C. Maranon, P. O. Vidalain, O. Azocar, D. Hanau, P. Lebon, M. Burgard, C. Rouzioux, V. Vullo, H. Yagita, et al
2004
. HIV type 1-infected dendritic cells induce apoptotic death in infected and uninfected primary CD4 T lymphocytes.
AIDS Res. Hum. Retroviruses
20
:
175
-182.
43
Chaperot, L., A. Blum, O. Manches, G. Lui, J. Angel, J. P. Molens, J. Plumas.
2006
. Virus or TLR agonists induce TRAIL-mediated cytotoxic activity of plasmacytoid dendritic cells.
J. Immunol.
176
:
248
-255.
44
Lu, G., B. M. Janjic, J. Janjic, T. L. Whiteside, W. J. Storkus, N. L. Vujanovic.
2002
. Innate direct anticancer effector function of human immature dendritic cells. II. Role of TNF, lymphotoxin-α1β2, Fas ligand, and TNF-related apoptosis-inducing ligand.
J. Immunol.
168
:
1831
-1839.
45
Korthals, M., N. Safaian, R. Kronenwett, D. Maihofer, M. Schott, C. Papewalis, E. Diaz Blanco, M. Winter, A. Czibere, R. Haas, et al
2007
. Monocyte derived dendritic cells generated by IFN-α acquire mature dendritic and natural killer cell properties as shown by gene expression analysis.
J. Transl Med.
5
:
46
46
De Panfilis, G., M. Venturini, A. Lavazza, M. A. Mommaas, D. Semenza, C. Torresani, G. Pasolini.
2003
. The tolerogenic molecule CD95-L is expressed on the plasma membrane of human activated, but not resting, Langerhans’ cells.
Exp. Dermatol.
12
:
692
-699.
47
Schmitz, M., S. Zhao, Y. Deuse, K. Schakel, R. Wehner, H. Wohner, K. Holig, F. Wienforth, A. Kiessling, M. Bornhauser, et al
2005
. Tumoricidal potential of native blood dendritic cells: direct tumor cell killing and activation of NK cell-mediated cytotoxicity.
J. Immunol.
174
:
4127
-4134.
48
Hardy, A. W., D. R. Graham, G. M. Shearer, J. P. Herbeuval.
2007
. HIV turns plasmacytoid dendritic cells (pDC) into TRAIL-expressing killer pDC and down-regulates HIV coreceptors by Toll-like receptor 7-induced IFN-α.
Proc. Natl. Acad. Sci. USA
104
:
17453
-17458.
49
Stary, G., C. Bangert, M. Tauber, R. Strohal, T. Kopp, G. Stingl.
2007
. Tumoricidal activity of TLR7/8-activated inflammatory dendritic cells.
J. Exp. Med.
204
:
1441
-1151.
50
Rissoan, M. C., T. Duhen, J. M. Bridon, N. Bendriss-Vermare, C. Peronne, B. de Saint Vis, F. Briere, E. E. Bates.
2002
. Subtractive hybridization reveals the expression of immunoglobulin-like transcript 7, Eph-B1, granzyme B, and 3 novel transcripts in human plasmacytoid dendritic cells.
Blood
100
:
3295
-3303.
51
Buzza, M. S., L. Zamurs, J. Sun, C. H. Bird, A. I. Smith, J. A. Trapani, C. J. Froelich, E. C. Nice, P. I. Bird.
2005
. Extracellular matrix remodeling by human granzyme B via cleavage of vitronectin, fibronectin, and laminin.
J. Biol. Chem.
280
:
23549
-23558.
52
Munn, D. H., M. D. Sharma, J. R. Lee, K. G. Jhaver, T. S. Johnson, D. B. Keskin, B. Marshall, P. Chandler, S. J. Antonia, R. Burgess, et al
2002
. Potential regulatory function of human dendritic cells expressing indoleamine 2,3-dioxygenase.
Science
297
:
1867
-1870.
53
Fuchs, A., M. Cella, T. Kondo, M. Colonna.
2005
. Paradoxic inhibition of human natural interferon-producing cells by the activating receptor NKp44.
Blood
106
:
2076
-2082.
54
Hanna, J., T. Gonen-Gross, J. Fitchett, T. Rowe, M. Daniels, T. I. Arnon, R. Gazit, A. Joseph, K. W. Schjetne, A. Steinle, et al
2004
. Novel APC-like properties of human NK cells directly regulate T cell activation.
J. Clin. Invest.
114
:
1612
-1623.
55
Kurabayashi, A., M. Furihata, M. Matsumoto, H. Hayashi, Y. Ohtsuki.
2004
. Distribution of tumor-infiltrating dendritic cells in human non-small cell lung carcinoma in relation to apoptosis.
Pathol Int.
54
:
302
-310.
56
Treilleux, I., J. Y. Blay, N. Bendriss-Vermare, I. Ray-Coquard, T. Bachelot, J. P. Guastalla, A. Bremond, S. Goddard, J. J. Pin, C. Barthelemy-Dubois, S. Lebecque.
2004
. Dendritic cell infiltration and prognosis of early stage breast cancer.
Clin. Cancer Res.
10
:
7466
-7474.
57
Munn, D. H., M. D. Sharma, D. Hou, B. Baban, J. R. Lee, S. J. Antonia, J. L. Messina, P. Chandler, P. A. Koni, A. L. Mellor.
2004
. Expression of indoleamine 2,3-dioxygenase by plasmacytoid dendritic cells in tumor-draining lymph nodes.
J. Clin. Invest.
114
:
280
-290.
58
Furset, G., Y. Floisand, M. Sioud.
2007
. Impaired expression of indoleamine 2, 3-dioxygenase in monocyte-derived dendritic cells in response to Toll-like receptor-7/8 ligands.
Immunology
123
:
263
-271.
59
Albert, M. L., B. Sauter, N. Bhardwaj.
1998
. Dendritic cells acquire antigen from apoptotic cells and induce class I- restricted CTLs.
Nature
392
:
86
-89.
60
Hoeffel, G., A. C. Ripoche, D. Matheoud, M. Nascimbeni, N. Escriou, P. Lebon, F. Heshmati, J. G. Guillet, M. Gannage, S. Caillat-Zucman, et al
2007
. Antigen crosspresentation by human plasmacytoid dendritic cells.
Immunity
27
:
481
-492.
61
Beaulieu, S., A. Kessous, D. Landry, S. Montplaisir, D. Bergeron, E. A. Cohen.
1996
. In vitro characterization of purified human thymic dendritic cells infected with human immunodeficiency virus type 1.
Virology
222
:
214
-226.
62
Beaulieu, S., M. Lafontaine, M. Richer, I. Courchesne, E. A. Cohen, D. Bergeron.
1998
. Characterization of the cytotoxic factor(s) released from thymic dendritic cells upon human immunodeficiency virus type 1 infection.
Virology
241
:
285
-297.
63
Raftery, M. J., M. Schwab, S. M. Eibert, Y. Samstag, H. Walczak, G. Schonrich.
2001
. Targeting the function of mature dendritic cells by human cytomegalovirus: a multilayered viral defense strategy.
Immunity
15
:
997
-1009.
64
Steinman, R. M., D. Hawiger, M. C. Nussenzweig.
2003
. Tolerogenic dendritic cells.
Annu. Rev. Immunol.
21
:
685
-711.
65
Huang, F. P., N. Platt, M. Wykes, J. R. Major, T. J. Powell, C. D. Jenkins, G. G. MacPherson.
2000
. A discrete subpopulation of dendritic cells transports apoptotic intestinal epithelial cells to T cell areas of mesenteric lymph nodes.
J. Exp. Med.
191
:
435
-444.
66
Baba, T., S. Iwasaki, T. Maruoka, A. Suzuki, U. Tomaru, H. Ikeda, T. Yoshiki, M. Kasahara, A. Ishizu.
2008
. Rat CD4+CD8+ macrophages kill tumor cells through an NKG2Dand granzyme/perforin-dependent mechanism.
J. Immunol.
180
:
2999
-3006.
67
Roncarolo, M. G., M. Bigler, J. B. Haanen, H. Yssel, R. Bacchetta, J. E. de Vries, H. Spits.
1991
. Natural killer cell clones can efficiently process and present protein antigens.
J. Immunol.
147
:
781
-787.
68
Wesa, A. K., W. J. Storkus.
2008
. Killer dendritic cells: mechanisms of action and therapeutic implications for cancer.
Cell Death Differ.
15
:
51
-57.