Langerhans cells and dendritic cells are cytotoxic towards HPV16 E6 and E7 expressing target cells

Abstract

Dendritic cells (DC) can be cytotoxic towards tumor cells by means of TNF family molecules expressed on the cell surface of activated DCs. Tumor cells expressing appropriate receptors are killed by DC, generating a source of antigen to be presented to the immune system. It has not been investigated whether Langerhans cells (LC) are selectively cytotoxic to tumor cells. This is of particular interest for epithelial tumor cells that physically interact with LC in vivo. Among epithelial tumors, the oncogenic process of cervical tumors is relatively well defined by their Human Papillomavirus (HPV) mediated etiology. To study whether HPV16 E6 and E7 expressions, otherwise observed in cervical tumor cells, can sensitize normal cervical epithelial cells to DC and LC mediated killing, the E6 and E7 genes were introduced by retroviral transfection, and cells were subsequently used as targets in cytotoxicity assays. Expression of cytotoxic molecules by effector cells was measured in response to the pro-inflammatory cytokine IFN-γ; cytotoxicity was established and concomitant expression of receptor molecules was assessed on target cells. A correlation between the shrinkage of HPV16 E6 and E7+ tumors versus DC and LC infiltration was evaluated in a murine model of cervical cancer. DC and LC proved to be equally cytotoxic towards E6 and E7 expressing cervical epithelial cells. IFN-γ induced TRAIL expression by DC and LC, and inhibition of TRAIL partially blocked cytotoxic effects. Expression of TRAIL decoy receptors was reduced following introduction of E6 and E7 into host cells. Shrinkage of HPV16 E6 and E7 expressing tumors correlated with infiltration by S100+ DC and LC, co-localizing with apoptotic mouse tumor cells. In conclusion, DC and LC mediated killing may be exploitable for anti-tumor treatment.

Introduction

Dendritic cells (DC) are key contributors to the immune response by processing local antigens, migrating to draining lymph nodes and presenting antigens to T cells and B cells [28]. DC can also be selectively cytotoxic towards malignantly transformed and/or virally infected cells [19]. Immature DC that actively phagocytize and process antigen reportedly induce tumor cell apoptosis by means of TNF family molecules expressed on the DC cell surface, provided the target cells express the appropriate receptors for these death ligands [23]. Resulting apoptotic target cells subsequently constitute a source of antigen to be processed and presented to T cells.

Exposure to cytokines including IFN-γ can reportedly enhance the cytotoxic ability of immature DC [9]. DC effector functions may be of particular importance in situations where target cells express reduced levels of MHC class I molecules, as reduced MHC expression provides an escape from immune surveillance by antigen restricted, CD8 T-cell mediated killing. Reduced MHC expression has been reported for tumor cells as well as for virus-infected cells [7, 26].

Human Papillomavirus (HPV) is among the viruses potentially sensitizing host cells to DC-mediated killing. In the skin, infection by low-risk HPV viruses can lead to the formation of warts [13]. Sexual transmission of high-risk HPV can lead to cervical cancer in women after the virus integrates into the host genome [36]. Selective infiltration of HPV infected cervical epithelium by cells expressing S100, a DC marker also expressed by Langerhans cells (LC), is suggestive of a potential contribution for DC or LC towards eliminating HPV infected cells, as supported by an association between a positive response to an HPV-directed cervical cancer vaccine and infiltration by S100+ cells [27]. Antibodies to S100 are routinely used to detect Langerhans cells in cervical neoplasia and have also recently served to validate antibodies to Langerin [27, 30].

Langerhans cells are a subset of DC that is found throughout epithelial compartments protecting underlying tissues from infection and cellular transformation [31]. LC differ from resident dermal dendritic cells by containing Birbeck granules involved in antigen processing [20]. It has been postulated that LC differ from dermal DC in their ability to process and present antigen [10, 15].

To investigate whether DC and LC can eliminate HPV16 E6 and E7-immortalized cells from the epithelium, it was investigated whether HPV16 E6 and E7 expressing cervical epithelial cells were susceptible to DC mediated cytotoxicity. This can shed light on the intracellular signaling pathways mediating expression of death receptors on the target cell membrane. E6 will bind and inactivate intracellular p53, whereas E7 binds Rb accelerating its degradation [29].

The cytotoxic activity of LC has not been reported to date, but tumor cells arising within the epithelium are primarily exposed to LC. Thus it was investigated whether LC are capable of killing cervical epithelial cells expressing the E6 and E7 genes of HPV16. LC can be generated in vitro from adherent blood monocytes by supplementing the cell culture media with GM-CSF, IL-4 and TGF-β [10]. Resulting LC were characterized regarding expression of Birbeck granules and E-cadherin, as well as of TNF-related apoptosis inducing ligand (TRAIL) previously implicated in DC mediated killing [10, 26]. Cytotoxicity was also correlated with the expression of TRAIL receptors by target cells in presence and absence of HPV16 E6 and E7. The potential involvement of TRAIL in DC or LC mediated killing was further established using soluble TRAIL and blocking antibodies to TRAIL.

Finally, the involvement of DC in tumor shrinkage and apoptosis of HPV+ cells was investigated in vivo. Mouse tumor models for cervical cancer include cell lines, such as C3, TC-1 and HLF16, the latter included in the current investigations expressing HPV16 E6 and E7 genes as well as activated ras [8, 11, 22]. This cell line was generated using fibroblasts isolated from heart and lung tissue of HLA-A2 D^d mice, transformed with an HPV16 E6 and E7 as well as an H-ras encoding expression vector. Resulting cell line HLF16 is derived from clonal cells demonstrating growth in soft agar [8]. The efficacy of anti-tumor vaccines can be assessed in these models by measuring a delay in tumor growth or accelerated tumor shrinkage [3]. The availability of mouse tumor models for cervical cancer has also enabled us to correlate tumor shrinkage and tumor cell apoptosis with infiltration by specific subgroups of immunocytes. An assessment was made on tumor cell apoptosis in proximity to infiltrating DC and LC using antibodies to S100. These combined in vitro and in vivo studies have enabled a detailed analysis of the potential contribution of DC and LC to shrinkage of HPV E6 and E7 expressing tissue, and of the cell surface molecules involved in this process.

Materials and methods

Cell culture

The DC and LC were generated from adherent human blood monocytes by allowing cells to adhere in AIMV medium (Invitrogen Corporation, Carlsbad, CA, USA) overnight. Non-adherent cells were removed, and cultures were maintained in AIMV supplemented with 100 ng/ml of GM-CSF (Berlex Laboratories Inc, Richmond, CA, USA), 25 ng/ml IL-4 (R&D Systems, Minneapolis, MN, USA) and antibiotics penicillin (100 IU/ml)/ streptomycin (100 μg/ml) and amphotericin (250 ng/ml) (Invitrogen). To generate LC, TGF-β (PeproTech Inc, Rocky Hill, NJ, USA) was added to this media at 10 ng/ml [10]. Media and cytokines were replenished after 3 days. On day 6, flow cytometry and cytochemistry of cytospins was used to confirm E-cadherin and langerin expression by LC compared to DC and to rule out contamination by lymphocytes beyond 10%. For detection of intracellular langerin expression by cultured human cells, cytospins were prepared from 10⁵ cells pre-incubated in paraformaldehyde 2% at RT for 10 min at RT, and then permeabilized with 0.03% saponin in PBS. Fixed cytospins were incubated with DCGM4 Ab to langerin (Immunotech). Biotinylated rabbit anti-mouse antiserum (Dako) was used in a second step and peroxidase-labeled streptavidin in the third staining step. Peroxidase activity was detected using amino ethyl carbazole (AEC) (Sigma) as a substrate.

Cells used in cytotoxicity assays were pre-exposed in part to IFN-γ at 1,000 U/ml (R&D Systems) for 48 h unless noted otherwise and washed before use.

Cervical epithelial cells were generated from otherwise discarded hysterectomy tissue incubated in presence of 0.25% trypsin O/N. These studies were approved by the Loyola University Medical Center IRB. Released cells were plated in Keratinocyte Growth Medium (Invitrogen) supplemented with 25 μg/ml bovine pituitary extract and 2.5 ng/ml recombinant EGF as directed (Invitrogen). The endocervical nature of resulting cell cultures was confirmed by immunostaining by antibodies to keratin 18 and absence of keratin 13 staining of adherent cells grown on Labtek multiwell glass slides (Nalge Nunc International, Naperville, IL, USA) [32]. Acetone fixed slides were incubated with 2DR antibody to keratin 13 (NeoMarkers, Fremont, CA, USA) or with DC10 antibody to keratin 18 (NeoMarkers), followed by peroxidase-labeled anti-mouse antiserum and incubation with AEC substrate.

Retroviral infection

Cervical epithelial cells were exposed to hexadimethrine bromide (polybrene) (Aldrich, Milwaukee, WI, USA) at 2 or 1 μg/ml, respectively for 4 h in combination with LXSN16E6E7 retrovirus containing supernatant from amphotropic producer cell line PA317 (American Type Culture Collection, Manassas, VA, USA). Cells were subsequently exposed to retrovirus without polybrene overnight. The procedure was repeated on day 3, and successful transfectants were selected in presence of geneticin (Invitrogen) 0.1 mg/ml for 8 days.

Cytotoxicity analysis and inhibition

Cytotoxicity was measured in routine JAM assays essentially as described [25]. Briefly, target cells were plated in 96 well plates and labeled with 1 μCi/well of tritiated thymidine (Amersham Corporation, Piscataway, NJ, USA) O/N. DC or LC were added in effector to target ratios of 4–10:1 in AIMV medium to part of the wells; AIMV medium only was added to control wells. Wells were carefully washed and ³H-thymidine retained by viable, adherent cells was measured after harvesting with a Packard Filtermate cell harvester (Perkin Elmer, Wellesley, MA, USA). Percent cytotoxicity was measured as 100[1 − (cpm with effectors − background)/(cpm without effectors − background)]. Statistically significant differences were determined by Student’s t test. The involvement of TRAIL in DC effector functions was assessed by performing the JAM assay in presence of DC 8:1 or soluble killer TRAIL (Qbiogene Incorporation, Carlsbad, CA, USA) at 200 ng/ml for 72 h in the presence or absence of sterile RIK-2 Ab to TRAIL at 5 μg/ml (eBioscience, San Diego, CA, USA) as described previously [2, 18].

Flow cytometry

Cells were scraped from culture dishes with a rubber policeman and incubated with primary antibodies. Primary antibodies used include M-301 to TNFα (Endogen, Woburn, MA, USA), rabbit polyclonal antiserum to human TRAIL (Calbiochem, San Diego, CA, USA), NOK-1 mouse monoclonal antibody to FasL (Caltag, Burlingame, CA, USA), MAB225 mouse monoclonal antibody to TNFRI (R&D Systems), MAB226 mouse monoclonal antibody to TNFRII (R&D Systems), mouse monoclonal antibody 32A1380 to TRAILR1 (Imgenex, San Diego, CA, USA), mouse monoclonal antibody 54B1005 to human TRAILR2 (Imgenex), Rabbit polyclonal antiserum to TRAILR3 (Orbigen Inc, San Diego, CA, USA), Goat polyclonal antiserum to TRAILR4 (R&D Systems) and MAB142 mouse monoclonal antibody to Fas (R&D Systems), HB15A mouse monoclonal to CD83 (Immunotech, Marseille, France), IT2.2 mouse monoclonal to CD86 (Pharmingen, San Diego, CA, USA), L243 mouse monoclonal to HLA-DR (Becton Dickinson, San Jose, CA, USA), CBR-p150/4G1 mouse monoclonal to CD11c (US Biological, Swampscott, MA, USA) or Cris-7 mouse monoclonal to CD3 (Cymbus biotechnology LTD, Chandlers Ford, UK) at empirically determined concentrations within the range provided by the manufacturer. Biotinylated, species-specific secondary antibodies were used in the second step (Dakopatts), followed by phycoerythrin-labeled streptavidin (Dakopatts). Fluorescence of 10,000 cells was determined by flow cytometry using an FACScalibur benchtop flow cytometer (BD biosciences, San Jose, CA, USA) equipped with a 15 mW argon-ion laser for detection of fluorescence plus right and forward angle scatter, with CellQuest software to control data acquisition.

Tissue immunohistology and immunocytology

Mouse tissue was obtained by resecting subcutaneous HLF16 tumors with known growth curves at the end of a tumor challenge experiment, with consent from the Loyola University Institutional Animal Care and Use Committee (IACUC). Data from this experiment were published previously [3]. Briefly, HLA-A^*0201 transgenic mice (n = 10/gp) received 2 × 10⁶ HLF16 tumor cells on day 0 and were vaccinated with viral replicon particles (VRP) at days 5, 10, and 15. Forty days following tumor challenge, mice vaccinated with a GFP encoding vaccine were 0% tumor free, whereas mice vaccinated with modified sequences derived from HPV16 E6 and E7 were between 50 and 100% tumor free, depending on the modifications introduced. For the purpose of this experiment, tumors from mice vaccinated with HPV16 derived sequences with four mutations within the E6 and E7 open reading frames were compared to GFP vaccinated mice. This E6E7 vaccinated group was 50% tumor free 40 days following tumor challenge, as represented in Fig. 5 of the earlier publication. Frozen mouse tumor sections were fixed in cold acetone and stained with the following biotinylated antibodies: 145-2C11 Armenian Hamster monoclonal to mouse CD3 (Pharmingen, San Diego, CA, USA), HL3 armenian hamster monoclonal to mouse CD11c (Pharmingen), C-20 goat polyclonal to mouse S100 (Santa Cruz Biotechnology), M-20 goat polyclonal to mouse CD14 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and T16 goat polyclonal to mouse CD68 (Santa Cruz Biotechnology).

Fig. 5.

TRAIL is involved in LC mediated killing of E6E7+ cervical epithelial cells. Cervical epithelial cells expressing HPV16 E6 and E7 were targeted by either soluble human killer TRAIL or by LC, and cytotoxicity was measured in a JAM assay performed in presence and absence of RIK-2 blocking antibodies to TRAIL, demonstrating a role for this TNF family member in LC mediated cytotoxicity

Single stainings were performed with secondary, isotype specific, peroxidase-labeled antibodies or streptavidin, using aminoethyl carbazole (AEC) as a substrate. These sections were counterstained with Harris hematoxilin (Sigma, St. Louis, MI, USA). For double stainings, alkaline phosphatase labeled isotype-specific antibodies were used in combination with isotype-specific peroxidase labeled antibodies. Fast Blue BB (Sigma) was the substrate to detect alkaline phosphatase activity [34].

Apoptotic cells were located by TUNEL staining according to the manufacturers’ instructions (Chemicon International, Temecula, CA, USA). Briefly, formalin-fixed sections were incubated with digoxygenin-labeled nucleotides and Klenow fragment of DNA polymerase I at 37°C. Single-stranded DNA fragments from apoptotic cells serve as a template for DNA synthesis. Incorporated label is detected by biotinylated anti-digoxygenin antibody and peroxidase-labeled streptavidin. For double stainings on cryosections, incubation with the alternate primary antibody was performed prior to formalin fixation unless noted otherwise.

Immunocyte infiltration of mouse tumors was independently assessed by two investigators as the number of stained cells within 10 randomly chosen optical fields. Statistically significant differences in infiltrating cell numbers were analyzed by Student’s t test.

Results

Cytotoxicity assays

The DC mediated cytotoxicity towards Ec0206 P4, Ec0315 P8 and Ec0408 P4 cervical epithelial cells from three donors expressing HPV16 E6 and E7 is demonstrated in Fig. 1, where DC cytotoxicity was clearly activated by IFNγ. In Fig. 2a, the LC phenotype of cultured cells was demonstrated. Elevated expression of E-cadherin as well as langerin by immature LC cells compared to immature DC was observed by FACS analysis and immunohistology, respectively. The percentage of cells expressing E-cadherin was 14% for immature DC (MFI 26) versus 92% for immature LC (MFI 558), respectively. Preferential langerin expression by cytospins of LC versus DC is confirmed in Fig. 2b. In Fig. 3 cytotoxicity towards cervical epithelial cells expressing both E6 and E7 is compared for DC and LC derived from the same donor. The 5.8% difference in average cytotoxicity between DC and LC was non-significant; thus the cytotoxic ability of LC is directly comparable to DC.

Fig. 1.

Dendritic cells (DC) cytotoxicity towards E6 and E7 expressing cervical epithelial cells is induced by IFN-γ. Epithelial cell cultures from three donors were transfected to express HPV encoded E6 or E7 genes targeted and were targeted by immature or IFN-γ treated dendritic cells (1,000 IU/ml for 48 h). The mean percent cytotoxicity (±SD) was measured in a JAM assay and results were displayed on the Y-axis

Fig. 2.

Langerhans cells are generated from adherent monocytes cultured in presence of GM-CSF, IL-4 and TGF-β. a Expression of E cadherin is elevated in presence of TGF-β compared to adherent monocytes exposed to dendritic cell culture conditions as measured by FACS analysis and b langerin expression is observed in cells exposed to TGF-β compared to adherent monocytes maintained under standard DC culture conditions

Fig. 3.

Langerhans cells are equally cytotoxic towards E6E7 transfected cervical epithelial cells compared to dendritic cells. E6E7 transfected cervical epithelial cells were radiolabeled and cocultured with or without Langerhans cells or dendritic cells to measure cytotoxicity in 72 h by a JAM assay

Flow cytometric analysis of membrane expression of TRAIL and its receptors

In Fig. 4 it is shown that membrane expression of TRAIL is upregulated following treatment of DC with IFN-γ, the MFI increasing 21-fold; whereas cells did not express TRAIL in the absence of IFN-γ, the percentage of TRAIL expressing cells following IFN-γ treatment was 54%. Upregulation of cell surface TRAIL expression by 49% of IFN-γ treated LC was similar to that observed for DC (not shown). Expression of TNF-α receptors was not induced by IFN-γ treatment in either DC or LC (not shown), supporting TRAIL as a more likely candidate molecule mediating cytotoxicity by IFN-γ treated DC and LC. In contrast to DC, some FasL expression was detectable in LC (1.6-fold over background) which was increased twofold in response to IFN-γ treatment of LC (not shown).

Fig. 4.

Membrane expression of TRAIL is elevated in presence of IFN-γ. DC treated with 1,000 U/ml of IFN-γ for 48 h were assessed for TRAIL expression by a FACS analysis and by b immunostaining of DC cytospins, together showing upregulated TRAIL expression located on the membrane of DC

Expression of TRAIL decoy receptor TRAILR3 was markedly downregulated in E6 and E7 expressing cells (MFI 238, CV 66) compared with untransfected parental cells (MFI 329, CV 85), reducing the mean fluorescence intensity by 28%. In separate experiments where expression of TRAILR3 was compared among skin keratinocytes expressing HPV16 E6 and E7 and parental cells from the same donor, a similar reduction in TRAILR3 expression was observed, whereas expression levels of TRAILR1, TRAILR2 and TRAILR4 were not affected by introducing the E6 and E7 genes of HPV16 (not shown).

TRAIL cytotoxicity towards E6 and E7 expressing epithelial cells

To further assess the contribution of TRAIL in LC mediated killing towards HPV16 E6 and E7 expressing cervical epithelial cells, cytotoxicity was measured in the presence of antibodies to TRAIL using E6 and E7 expressing cervical epithelial cells as target cells as illustrated in Fig. 5. As a control, cytotoxicity was assessed in presence of soluble killer TRAIL (71 ± 7%) and inhibited by antibodies to TRAIL (to 46 ± 18%). The cytotoxicity towards cervical epithelial cells transfected to express HPV16 E6 and E7 genes was significantly reduced in the presence of antibodies to TRAIL (from73 ± 9 to 55 ± 9% in 72 h), confirming a role for TRAIL in LC mediated target cell death. These data further support a role for TRAIL in LC mediated killing of E6 and E7 expressing cervical epithelial cells.

Tissue immunohistology

Figure 6 shows a highly significant increase in the abundance of S100 expression observed only in tumors from mice treated with an effective HPV16 E6 and E7 encoding VRP vaccine, and not in tumors from mice treated with an irrelevant control vaccine (encoding green fluorescent protein). No other significant differences were observed among tumors from ineffectively and effectively treated mice, indicating that the abundance of DC and LC, rather than T cells or macrophages correlated with treatment efficacy. Interestingly, S100 expressing cells were particularly abundant in areas of massive apoptosis within HLF16 tumors as demonstrated in Fig. 7, and such co-localization further implicates DC and LC in mediating tumor cell death.

Fig. 6.

Tumor shrinkage is associated with infiltration by S100+ immunocytes. Resected HPV16 E6E7+ tumors growing in mice treated with viral replicon particles encoding the E6 and E7 genes or an irrelevant gene (GFP) were assessed for infiltration by several immunocyte subsets, demonstrating that tumor shrinkage is markedly associated with infiltration by S100+ DC and LC

Fig. 7.

Apoptotic mouse tumor cells are found in close proximity to S100+ dendritic cells. Areas of increased TUNEL positivity containing apoptotic tumor cells were infiltrated by elevated numbers of S100+ DC and LC. Such S100+ cells were frequently found in close proximity to dying tumor cells. Scale bar in enlarged image: 20 μm

Discussion

Data presented currently support the notion that activated TRAIL expressing DC and LC contributes to the elimination of HPV infected cells from cervical epithelium. A role for TRAIL in LC mediated cytotoxicity was supported by the observed killing of E6 and E7 expressing cervical epithelial cells in presence of soluble killer TRAIL and partial inhibition of LC mediated killing by antibodies to TRAIL. The incomplete inhibition of cytotoxicity by anti-TRAIL antibodies can be explained in part by high expression of TRAIL on the LC cell surface exceeding the concentration blocked by the presence of antibodies in the assay. In the absence of functional assays to block FasL expressed by LC in response to IFN-γ exposure, an additional role for this ligand in LC mediated killing cannot be ruled out at this time. However, expression of HPV16 E6 and E7 has been shown to protect keratinocytes from FasL-induced apoptosis, correlating with reduced expression of Fas [1]. Importantly, in earlier studies we have observed that expression of HPV16 E6 and E7 by cervical epithelial cells sensitized host cells to DC mediated cytotoxicity. In those experiments the E6 and E7 genes were retrovirally introduced, either alone or in combination, into primary endocervical epithelial cells. When exposed to IFN-γ treated DC, the observed % cytotoxicity were 0.1 ± 5% for untransduced cells versus 27.8 ± 2.8% for E6 expressing targets, 21.2 ± 11/5% for E7 expressing targets and 32.8 ± 7.2% for E6 and E7 expressing endocervical epithelial cells in three representative experiments performed (unpublished data). Thus the HPV16 E6 and E7 genes each sensitize cervical epithelial cells to DC mediated cytotoxicity.

Host cells expressing the E6 and E7 genes exhibited reduced levels of TRAIL decoy receptor TRAILR3. These data can explain the enhanced sensitivity of HPV16 E6 and E7 expressing epithelial cells to TRAIL mediated apoptosis observed in our cytotoxicity assays, as in the absence of decoy receptors TRAIL can engage TRAILR1 and TRAILR2 instead, which carry death domains that activate the caspase mediated apoptotic pathway [5]. Known target molecules of the E6 and E7 gene products include p53 and pRb, and p53 has been shown to affect the expression of several TRAIL receptors [12]. However, in separate experiments reduced TRAILR3 expression was also observed in skin keratinocytes expressing only the E7 gene of HPV16 (unpublished observation), suggesting that pRb may similarly affect expression of TRAIL decoy receptors.

By affecting the host cell cycle, the E6 and E7 gene products contribute to cellular immortalization, yet expression of E6 and E7 per se is insufficient for complete malignant transformation of host cells. Thus DC and LC mediated cytotoxicity can likely target cervical epithelial cells that are not yet fully transformed. Based on their location, it can be speculated that in particular LC mediated cytotoxicity occurs as a natural part of the host defense to prevent cervical tumor formation, which will frequently go unnoticed. It is possible that activated host DC recruited to a tumor site following vaccination can similarly participate in an effective anti-tumor response. This notion is supported by the demonstrated efficacy of GM-CSF and Flt3L vaccines for tumor patients, although the consequences of this treatment for DC mediated cytotoxicity cannot be easily distinguished from its effect on the recruitment of tumor-specific T cells [4, 24]. Similarly, the anti-tumor efficacy of IFNγ treatment as assessed in clinical trials [19] may be regarded as supportive for a role of DC and LC mediated cytotoxicity, as the cytotoxic activity of DC and LC was enhanced after exposure of these cells to IFN-γ. It was demonstrated that exposure to IFN-γ was associated with elevated expression of TRAIL, extending data reported by others to LC [9]. In fact, plasmacytoid DC reportedly express interferon in response to viral or bacterial exposure, due to expression of TLR7 and TLR9 [6]. Sensitivity of HPV-positive cervical cells to TRAIL-induced apoptosis is similarly supported by recently reported data [14].

Regarding the relevance of using cultured cervical epithelial cells for our studies, cervical epithelial cells isolated from otherwise discarded hysterectomy tissue expressed keratin 18 and lack keratin 13 expression, suggesting that the culture conditions favor outgrowth of endocervical cells [32]. HPV is known to infect cells at the squamo-columnar junction followed by endocervical and ectocervical epithelium in decreasing order of frequency [36]. Although malignantly transformed cervical epithelial cells can extend towards the ectocervix, endocervical tumors are also observed indicating that cultured cytokeratin18+ cells are relevant cells to host the HPV16 encoded genes [33].

By inducing an LC-like phenotype in adherent monocytes through cultivation in presence of GM-CSF, IL-4 and TGF-β, we were able to demonstrate that the LC subfamily of dendritic cells is cytotoxic towards E6 and E7 expressing cervical epithelial cells. This is of particular relevance because the localization in vivo dictates that LC are possibly the first subset of dendritic cells to encounter HPV infected epithelial cells in vivo. LC effector functions have not previously been reported. DC effector functions were previously shown to be effective towards a wide array of tumor cells, leaving normal tissue cells untouched [16]. Thus it is likely that LC can effectively target diverse epithelial tumors including basal cell carcinomas, squamous cell carcinomas, melanomas and mucosal tumors of the digestive tract and contribute to tumor shrinkage. Stimulating LC infiltration of epithelial tumors and enhancing expression of TNF family molecules, in particular TRAIL expression, may thus offer an effective means to treat epithelial tumors. This is supported by protective anti-tumor immunity reportedly induced by in situ Langerhans cell vaccination [21].

Mouse models of cervical cancer offer the unique opportunity to quantify the efficacy of anti-tumor vaccines against a homogeneous genetic and physiologic background. In the current setting, tumor tissue harvested following vaccination with effective versus control vaccines clearly demonstrated an association between vaccine efficacy and tumor infiltration by S100+ DC and LC, but not by T cells or other immunocytes. This is supportive of the notion that it is possible to induce tumor shrinkage supported by cytotoxic DC and LC, particularly because apoptotic tumor cells clearly colocalized with infiltrating S100+ cells. Antibodies to mouse langerin potentially suitable for mouse tissue immunohistology are currently unavailable. Such antibodies may eventually be used to assess infiltration of mouse tumors by LC and colocalization of mouse LC and tumor cells.

An association between HPV infection and LC infiltration is also supported by findings reported by Jimenez-Flores et al. [17]elegantly demonstrating that LC morphology and distribution is altered by the presence of HPV in epithelial sheets from cervical tissue.

Taken together, DC and LC can be cytotoxic towards HPV-derived E6 and E7 expressing immortalized, yet not necessarily fully transformed epithelial cells. Cytotoxicity is mediated at least in part by TRAIL expression, which can be exploited to prevent outgrowth of malignantly transformed epithelial cells by enhancing tissue infiltration and expression of TRAIL by DC and LC, for example by IFNγ exposure. It will be of interest to assess the contribution of additional factors known to affect DC activation towards cytotoxic effector functions, including dsRNA and CD40L [35].

Abbreviations

DC

Dendritic cells

HPV

Human Papillomavirus

IFN-γ

Interferon-gamma

LC

Langerhans cells

TGF-β

Transforming growth factor-beta

TNF

Tumor necrosis factor

TRAIL

TNF- related apoptosis-inducing ligand