HMGB1 secretion during cervical ... - Wiley Online Library

IJC International Journal of Cancer

HMGB1 secretion during cervical carcinogenesis promotes the acquisition of a tolerogenic functionality by plasmacytoid dendritic cells phanie Demoulin1, Michael Herfs1, Joan Somja2, Patrick Roncarati1, Philippe Delvenne2 and Pascale Hubert1 Ste 1 2

Department of Pathology, Laboratory of Experimental Pathology, University of Lie`ge, GIGA-Cancer, 4000 Lie`ge, Belgium Department of Pathology, University Hospital of Lie`ge, 4000 Lie`ge, Belgium

Despite the evidence that human oncogenic papillomavirus (HPV) are strongly implicated as the causative agents in the etiology of cancers such as cervical, vulvar or anal cancer and

Key words: plasmacytoid dendritic cells, tolerogenicity, Treg cells, cervical cancers, HMGB1 Abbreviations: CpG ODN: CpG oligonucleotides; EpM: epithelial metaplasia; HMGB1: high-mobility group box 1; HPC: hematopoietic progenitor cells; HPV: human papillomavirus; HSIL: highgrade squamous intraepithelial lesions; IFN: interferon; IHC: immunohistochemistry; LSIL: low-grade squamous intraepithelial lesion; MFI: mean fluorescence intensity; NK: natural killer cell; pDCs: plasmacytoid dendritic cells; SCC: squamous cell carcinoma; SIL: squamous intraepithelial neoplasia; TLR: toll-like receptor; Treg cells: regulatory T cells Additional Supporting Information may be found in the online version of this article. Grant sponsor: Belgian Fund for Medical Scientific Research; Fonds Leon Fredericq DOI: 10.1002/ijc.29389 History: Received 4 June 2014; Accepted 26 Nov 2014; Online 9 Dec 2014 Correspondence to: Pascale Hubert, Laboratory of Experimental Pathology (GIGA-CANCER), ULg, Sart Tilman, Tour 3 14, B-4000 Lie`ge, Belgium, Tel.: 132-4-366-42-82, Fax: 132-4-366-95-83, Email: [email protected]

C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

their precursors, HPV infection alone is not sufficient for cancer development. The intrinsic immunity plays a key role in controlling HPV infection and the subsequent development of squamous intraepithelial lesions (SIL), as shown indirectly by the increased frequency of HPV-associated lesions in patients with depressed cell-mediated immunity.1 Increasing evidence show that upregulation of immunosuppressive factors (e.g., PGE2) and deregulation of soluble factors (e.g., TGF-b, TNFa, MIP3-a or IL-10) expression represent key steps in cervical cancer progression.2 For example, deregulation of soluble factors expression such as MIP3a and TGF-b was notably reported to reduce the density of antigen presenting cells3,4 that, in other circumstances, would stimulate the antitumor adaptive immunity. Constituting a subpopulation of dendritic cells, plasmacytoid dendritic cells (pDCs) are a rare cell type specialized in the secretion of large amounts of type I interferon (IFN) in response to pathogens or motifs such as synthetic CpG oligonucleotides (ODN). Their ability to mount a robust type I IFN response in acute viral infections is linked to their selective expression of toll-like receptors 7 and 9 (TLR7 and 9).5 Although, pDCs infiltration has already been reported in several types of cancer (head and neck cancer, ovarian carcinoma, breast cancer, lung cancer), their implication in antitumor response is largely debated.6 Some studies have reported that pDCs could limit tumoral progression by secreting type I IFN, cytokines known for their antitumor role. Indeed,

Tumor Immunology

Acquisition of an impaired functionality by plasmacytoid dendritic cells (pDCs) contributing to cancer progression has been documented in different types of cancers. In the present study, we postulate that molecules secreted by (pre)neoplastic epithelial cells of the genital tract (cervix/vulva) might attract pDCs but also modify their proper functionality, allowing these cells to initiate a tolerogenic response interfering with antitumor immunity. We demonstrated that pDCs are recruited during the cervical metaplasia-dysplasia-cancer sequence, through the action of their chemoattractant, chemerin. We showed that stimulated-pDCs exposed to cervical/vulvar tumor microenvironment display an altered phenotype. We also demonstrated that cervical/vulvar neoplastic keratinocytes inhibit the proper function of pDCs by decreasing their IFNa secretion in response to CpG oligonucleotides. In parallel, we observed that (pre)neoplastic areas of the cervix are infiltrated by FoxP31 Treg cells which colocalize with pDCs. Accordingly, pDCs cocultured with cervical/vulvar neoplastic keratinocytes have the capacity to induce a Treg cell differentiation from na€ıve CD41 T cells, which is in agreement with the development of a tolerogenic response. We identified HMGB1 as a soluble factor produced by neoplastic keratinocytes from the genital tract involved in pDCs functional alteration. Indeed, this molecule inhibited pDC maturation, decreased IFNa secretion following TLR9 stimulation and forced these cells to become tolerogenic. In contrast, inhibition of HMGB1 restored pDC phenotype. Our findings indicate that the use of inhibitory molecules notably directed against HMGB1 in cervical/vulvar (pre)neoplastic lesions might prevent alterations of pDCs functionality and represent an attractive therapeutic strategy to overcome immune tolerance in cancers.

346

Cervical HMGB1 induces tolerogenic functionality in tumor-associated pDCs

Tumor Immunology

What’s new? The human immune system usually strikes with deadly efficiency. Somehow, though, tumors evade destruction, despite the presence of distinctive tumor antigens. In fact, tumors actively train the immune cells to tolerate their antigens, using plasmacytoid dendritic cells (pDCs). In this paper, the authors asked whether cervical cells send out a molecular signal that summons the pDCs and enlists them in the service of the tumor. They discovered that the protein HMGB1, produced by tumor cells, interfered with pDCs’ maturation and created an immune tolerance toward the tumor. Inhibiting HMGB1 restored normal pDC function, suggesting a possible treatment angle.

IFNa presents direct antitumor activities by inhibiting tumor cell proliferation, (lymph)angiogenesis and tumor metastasis but also by promoting immunosurveillance through the activation of B cells, NK cells and macrophages.7–10 Furthermore, given their capacity to bring together innate and adaptive immune responses, pDCs may be important players in the tumor microenvironment. Nevertheless, emerging evidence also suggest that pDCs could play a negative regulatory role in antitumor immunity.6 Indeed, pDCs recruited to the tumor microenvironment often display an immature phenotype and may contribute to immune tolerance by promoting regulatory T cells (Treg) differentiation or expansion.11 Other pejoratives roles of pDCs in tumor pathology have also been described such as their implication in tumor angiogenesis via the production of TNF-a and IL-812 and in bone metastasis of breast cancer cells.13 pDCs involvement in tumor immunity was shown to have a clinical impact because their infiltration in tumors such as breast and ovarian cancers, and melanoma was recently associated with poor clinical outcomes.14–16 It has been postulated that tumors may escape the immune system by an impairment of pDCs functionality mediated through the local production of immunosuppressive cytokines.6 Squamous cell carcinoma (SCC) of the genital tract is known to secrete immunosuppressive factors, but their effects on pDCs functionality are still unknown. First described as a DNA chaperone, high-mobility group box1 (HMGB1) was then identified as an alarmin, a protein presenting crucial functions to alert and mobilize the immune response during infection and tissue damage. Indeed, in response to injury, infection and other inflammatory stimuli, HMGB1 can be released in the extracellular space either passively from necrotic cells17 or actively by activated macrophages, neutrophils, mature dendritic cells and NK cells.18,19 HMGB1 has also been proposed to be a crucial mediator in the pathogenesis of cancer. In the tumor setting, HMGB1 can also be secreted directly by cancer cells. Not only is overproduction of HMGB1 associated with the hallmarks of cancer (e.g., tumor (lymph)angiogenesis, growth, inflammation, invasion, and metastasis)20 but recent data also showed that this molecule could inhibit antitumor immunity notably by enhancing IL-10 production by Treg cells and by inhibiting monocyte-derived dendritic cells function.21,22 In particular, recent data reported that human recombinant HMGB1 suppress pDCs proinflammatory cytokine secretion

(IFNa, IL-6, TNF-a, IP-10) and maturation in response to TLR9 agonists. In addition, HMGB1 is able to prevent the upregulation of costimulatory and adhesion molecules on pDCs and suppresses their ability to drive the generation of IFNg-secreting T cells.23 Interestingly, in women presenting SCC of the uterine cervix, high serum levels of HMGB1 were inversely correlated with disease-free and overall survival.24 In this study, we tested the hypothesis that impaired antitumor immunity in cervical/vulvar (pre)neoplastic lesions could promote tumor progression. We evaluated the differential density of pDCs and Treg cells in the normal, metaplastic and (pre)neoplastic cervical lesions. We also determined the expression of chemerin, a chemoattractant linked to pDCs recruitment in lymphoid organs, inflamed skins and tumors.25 Moreover, we carried out in vitro studies to determine the activation status and functionality of pDCs in a tumoral context by using SCC cell lines from the genital tract. HMGB1 emerged as an immunosuppressive factor acting on pDCs, through their receptor RAGE, to render them tolerogenic during cervical/vulvar carcinogenesis. Therefore, inhibition of HMGB1 activities within SCC of the genital tract could emerge as an appealing therapeutic strategy to trigger antitumor immunity.

Material and Methods Tissue specimens

Paraffin-embedded cervical or vulvar specimens were retrieved from the Tumor Biobank of the University Hospital of Lie`ge. Cervical tissue samples included normal exocervical tissues, epithelial metaplasia (EpM), low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion (HSIL) and squamous cell carcinoma (SCC). Vulvar tissue samples included high-grade intraepithelial neoplasia (VIN II and VIN III). The protocol was approved by the Lie`ge University Hospital Ethics Committee. Immunohistochemistry

Serial sections of cervical biopsy specimens underwent immunostaining using antibodies directed against BDCA-2 (clone 124B3.13; Dendritics, Lyon, France), chemerin (Phoenix Pharmaceuticals, CA), FoxP3 (clone 236A/E7; eBioscience, CA) and HMGB1 (clone 1D5; Abnova, Taiwan) (see Supporting Information Table). Vulvar biopsy specimens also underwent immunostaining for HMGB1. Sections were incubated with the primary antibodies either overnight at 4 C C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

347

Demoulin et al.

Generation of pDCs from CD341 hematopoietic progenitor cells

pDCs represent a rare cell population and it is historically challenging to collect enough cells per donor (in the peripheral blood or in tissue specimens) to perform the different assays. To resolve this technical problem, pDCs were generated from human cord blood CD341 hematopoietic progenitor cells (HPC) cultured for 21 days in the presence of TPO (10 ng ml21), Flt3L (100 ng ml21) and IL-3 (20 ng ml21) (Peprotech, Rocky Hill, NJ) by using a previously described protocol.27 All human samples were collected according to a protocol approved by Lie`ge University Hospital Ethics Committee. Cocultures of pDCs in the presence of immortalized skin keratinocytes or cervical/vulvar cell lines

SiHa and CaSki are cervical SCC cell lines infected by HPV16. A431 is a vulvar carcinoma cell line negative for HPV DNA. HaCat are immortalized normal skin keratinocytes. All cell lines were obtained from the ATCC (American Type Culture Collection, Manassas, VA). Cell cultures were maintained following ATCC reported methods. Cocultures were initiated by seeding either HaCat cells or cervical/vulvar malignant keratinocytes (SiHa, CaSki, A431) (5 3 104 cells) on 0.4-mm pore size membrane inserts (Nunc, Roskilde, Denmark) in a six-well plate into which 106 pDCs were seeded. At 24 hr after the coculture, inserts were extracted and 12 mg ml21 of CpG ODN (CpG ODN type A: 5’-ggGGGACGATCGTCgggggg-3’; Eurogentec, Seraing, Belgium) or 0.5 mg ml21 of R848 (InvivoGen, San Diego, CA) were added to the wells in order to induce pDCs maturation. In selected experiments, HMGB1 inhibitors or a neutralizing anti-RAGE antibody were added [(monoclonal anti-HMGB1 C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

antibody (DPH1.1 mAb) (0.5 mg ml21; HMGBiotech, Milano, Italy); A box peptide (1.25 mg ml21; HMGBiotech); mouse monoclonal anti-RAGE antibody (1.25 mg ml21; Abcam)]. Data from cocultures between pDCs and SCC cell lines were compared to a control corresponding to pDCs cultured alone (=100%). In contrast, data from experiments using HMGB1 inhibitors or an anti-RAGE antibody were compared to a control corresponding to pDCs cultured alone but also incubated with either an HMGB1 inhibitor (DPH1.1, A box peptide) or an anti-RAGE Ab (=100%). Culture of pDCs in the presence of human recombinant PGE2 or HMGB1

pDCs were incubated with medium alone or different concentrations of PGE2 (1, 10, and 100 mM) (Cayman Chemical, USA) or HMGB1 (0.05, 0.25, and 2 mg ml21) (R&D Systems, Minneapolis, MN) 24 hr before stimulation with CpG ODN. In selected experiments, HMGB1 inhibitors or a neutralizing anti-RAGE antibody were added in the culture system [monoclonal anti-HMGB1 antibody (DPH1.1 mAb) (4 mg ml21; HMGBiotech); A box peptide (10 mg ml21; HMGBiotech); quercetin (50 or 100 mM; Sigma–Aldrich, StLouis, MO); mouse monoclonal anti-RAGE antibody (10 mg ml21; Abcam)]. pDCs cultured alone or in the presence of HMGB1 inhibitors or the neutralizing anti-RAGE antibody were used as controls (=100%). Annexin V-PI assay

To assess the percentage of apoptotic/necrotic cells, our cell lines were incubated with annexin V-FITC and propidium iodide (PI) according to the manufacturer’s recommendations (BD Biosciences). Results were collected by flow cytometry (BD FACSCalibur flow cytometer, BD Biosciences). Flow cytometric analysis

Mouse monoclonal anti-human antibodies (CD123-FITC, CD11c-APC, BDCA-4-PE, CD40-PE, CD83-PE, CD86-PE, CCR7-PE, HLA-DR-PE, ICOSL-PE and RAGE) were used in flow cytometric analysis (see Supporting Information Table). RAGE antibody was coupled with an anti-mouse NL637-conjugated antibody. Mean fluorescence intensity and positive cell percentages were measured on a BD FACSCanto II flow cytometer (Becton Dickinson, NJ) and data were analyzed using BD FACSDiva software V 6.1.2 (Becton Dickinson). Flow plots were generated by using the flow cytometry data analysis Software, FlowJo (Ashland, OR). Cytokine production assays

Culture supernatants collected from pDCs cultures were assessed for IFNa levels by using ELISA kits according to the manufacturer’s instructions (PBL InterferonSource, NJ). HMGB1 concentrations were assessed in HaCat, SiHa, CaSki and A431 cell lines by using an ELISA kit according to the manufacturer’s instructions (Gentaur, Kampenhout, Belgium). All assays were performed using duplicate samples.

Tumor Immunology

(BDCA-2, HMGB1) or for 1 hr (FoxP3) or 2 hr (chemerin) at room temperature. The revelation was performed with the CSA II kit (Dako, Glostrup, Denmark) (FoxP3), with the peroxidase LSAB2 system (DAKO) (BDCA-2 and chemerin) or with the Envision detection kit (Dako) (HMGB1), according to the suppliers’ recommendations. Chemerin and HMGB1 immunostainings were evaluated by using a semiquantitative score of the intensity and extent of the staining according to an arbitrary scale.26 Double immunostainings with FoxP3 and BDCA-2 were performed to analyze the colocalization of Treg cells with pDCs by using an automated immunostainer (Ventana Medical Systems, Tucson, AZ). HMGB1 expression was also assessed by immunofluorescence by using an antiHMGB1 antibody (clone 1D5; Abnova) coupled with a secondary antibody (goat anti-mouse Alexa Fluor 488, Invitrogen) (see Supporting Information Table). Nuclei were counterstained with the TOTO-3 iodide dye (Molecular Probes, Leiden, The Netherlands) for 20 min at ambient temperature. Immunofluorescence patterns were visualized using a confocal image analysis system (Leica TCS SP2 Laser scanning spectral confocal microscope, Heidelberg GmBH, Germany) equipped with a 633 objective.

348


Boyden chamber assay

pDCs migration was evaluated using a chemotaxis microchamber technique (48-well Boyden microchamber; Neuroprobe, Cabin John, MD) as previously described.27 Different concentrations of human recombinant chemerin (10 pM, 100 pM, 10 nM or 100 nM; R&D Systems) were used. Conditioned medium of human fibroblasts was used as positive control.

Tumor Immunology

Treg cell induction assay

The assay was performed by culturing pDCs (stimulator cells) isolated from coculture experiments with allogeneic CD41 T cells (responder cells) sorted from peripheral blood mononuclear cells using the MACS CD41 T Cell Isolation Kit (Miltenyi Biotec GmBH, Bergisch Gladbach, Germany), according to the manufacturer’s protocol. The stimulator-toresponder ratio corresponded to 1:10, and cells were placed in RPMI 5% human pooled AB serum (Invitrogen, Carlsbad, CA) in six-well plates (Nunc) for 7 days. pDCs cultured alone were used as control. FoxP3 expression in natural Treg cells could not be used as a control for our experiments. Indeed, in agreement with recent data,28,29 flow cytometry analysis and real-time PCR show that those cells express significantly less FoxP3 compared to induced Treg cells obtained by the culture of na€ıve CD41 T cells with nonactivated pDCs (Supporting Information Fig. 1). Natural Treg cells (CD41, CD251, CD127low or neg) were sorted from blood using a BD FACSAria III sorter. After the Treg cell induction assay, cells were harvested and FoxP3 expression was assessed by qPCR or by flow cytometry analysis. For flow cytometry analyses, FoxP3 expression was assessed on CD41 CD251 CD127low or neg T cells. After cell surface markers staining (CD4-PerCP, CD25BV421, CD127-PECy7) (see Supporting Information Table), FoxP3 staining was performed by using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Quantitative TLR9 and FoxP3 RT-PCR

Total RNA extracted from cell cultures (Nucleospin RNA II, Macherey-Nagel, France) was reverse-transcribed using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. Quantitative RT-PCR was then performed using FastStart SYBR Green Master (Roche Applied Science, Germany). PCR primers sequences were as follow: TLR9 F: 50 -CCC-GCT-ACT-GGT-GCT-ATC-C-30 ; TLR9 R: 50 -CCT-TCC-TCT-TTC-CAC-TCC-C-30 ; FOXP3 F: 50 -CAG-CAC-ATT-CCC-AGA-GTT-CCT-C-30 ; FOXP3 R: 50 -GCG-TGT-GAA-CCA-GTG-GTA-GAT-C-30 ; GAPDH F: 50 -ACC-AGG-TGG-TCT-CCT-CTG-AC-30 ; GAPDH R: 50 TGC-TGT-AGC-CAA-ATT-CGT-TG-30 (Eurogentec). All the experiments were performed in duplicate using the ABIPrism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) and negative controls (master mix without any cDNA) were added in each run. The results were ana-

lyzed by the comparative threshold cycle method (Ct) values and reference gene Ct values and normalized by GAPDH as an internal control. Statistical analysis

Statistical analyses were performed by using GraphPad Prism 4.00 (Graph-Pad Software, CA). Unpaired t test or Mann– Whitney test were performed when appropriate. Differences were considered as statistically significant when p < 0.05.

Results pDCs density and chemerin expression are increased along the cervical “metaplasia-dysplasia-cancer” sequence

We first investigated by immunohistochemistry (IHC), pDCs infiltration in normal exocervical tissues, epithelial metaplasia (EpM) and (pre)neoplastic lesions of the uterine cervix. Representative IHC staining examples are presented in Figure 1. The normal exocervical epithelium was poorly infiltrated by BDCA-21 pDCs (Fig. 1a). In contrast, the density of these cells was significantly higher in the epithelium associated to EpM (p < 0.01), LSIL (p < 0.001), HSIL (p < 0.01) and SCC (p < 0.001) (Fig. 1c). The BDCA-2 score was also significantly higher in the stromal tissue associated with EpM (p < 0.05), LSIL (p < 0.001), HSIL (p < 0.001) and SCC (p < 0.001) in the exocervix (Fig. 1c). pDCs have been shown to express the functional receptor for the chemerin, ChemR23 (CMKLR1). On the basis of those data, we sought to determine by IHC whether chemerin could be involved in pDCs recruitment in EpM and (pre)neoplastic lesions of the uterine cervix. Chemerin expression in normal exocervical epithelia was low (Fig. 1b). In contrast, the epithelia of EpM (p < 0.001), LSIL (p < 0.001), HSIL (p < 0.001) and SCC (p < 0.01) produced significantly higher levels of chemerin than the normal exocervix as illustrated in Figure 1d. In SCC, a heterogenic staining was frequently observed. These data suggest that the increased expression of chemerin could explain the recruitment of pDCs observed in cervical tissues. In agreement with this hypothesis, in a Boyden chamber assay, chemerin exhibited a significant chemotactic activity on pDCs. This chemoattraction was observed at a concentration as low as 100 pM (Fig. 1e). pDCs exposed to cervical/vulvar transformed keratinocytes display an altered phenotype and tolerogenic functionalities

We assessed the effects of soluble factors secreted by transformed keratinocytes of the genital tract (cervix/vulva) on pDCs phenotype and functionality by performing coculture assays with transwell membranes avoiding cell-to-cell contact. pDCs were either cultured alone (control pDCs), in the presence of immortalized normal skin keratinocytes (HaCat) or with SCC cell lines (SiHa, A431, CaSki) and then stimulated for maturation with a TLR9 (CpG ODN) or a TLR7 ligand (R848). pDCs exposed to these SCC cell lines and stimulated C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

349

Tumor Immunology

Demoulin et al.

Figure 1. pDCs density and chemerin expression are increased in both epithelial metaplasia and (pre)neoplastic lesions. (a) pDCs infiltration in cervical biopsy specimens was assessed by immunohistochemistry. To identify pDCs in tissue specimens, an anti-BDCA-2 antibody was used. Original magnifications: 3200. (b) Chemerin expression in paraffin-embedded sections was assessed by immunohistochemistry. Magnification: 3100. Note the strong anti-chemerin immunostaining in the epithelial cells and the slight and diffuse expression in the stromal compartment. (c) Semiquantitative evaluation of BDCA-21 pDCs density in the epithelia of the exocervix (EXO) (n 5 17), EpM (n 5 15), LSIL (n 5 14), HSIL (n 5 18) and SCC (n 5 15) or in associated stromal tissues. We observed an increased density of pDCs in the stromal compartment along the cervical “metaplasia-dysplasia-cancer” sequence. Data represent mean number of BDCA21 cells 6 SD in five high power (3400) fields of the microscope (*p < 0.05; **p < 0.01; ***p < 0.001). (d) Semiquantitative evaluation of chemerin expression in normal exocervical epithelium (EXO) (n 5 20), EpM (n 5 22), LSIL (n 5 19), HSIL (n 5 23) and SCC (n 5 18). Asterisks indicate statistically significant differences (**p < 0.01; ***p < 0.001). (e) Influence of human recombinant chemerin on pDCs migration in a Boyden chamber assay. HFM: conditioned medium of human fibroblasts (positive control). A non-conditioned (NC) medium was supplemented with four different concentrations (10 pM, 100 pM, 10 nM and 100 nM) of recombinant chemerin. Data represent mean 6 SD (*p < 0.05; ***p < 0.001) of three independent experiments. C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

Tumor Immunology

350


Figure 2. Molecules secreted by cervical/vulvar SCC cells alter pDCs phenotype and function. (a) pDCs were incubated for 24 hr either with immortalized normal skin keratinocytes (HaCat), cervical/vulvar transformed keratinocytes (SiHa, CaSki or A431) or alone (control). pDCs were then stimulated with CpG ODN A for 24 hr and expression of cell-surface molecules was measured by flow cytometric analysis. Data represent relative MFI of positive cells 6 SD of 12 independent experiments (*p < 0.05; ***p < 0.001). All data were normalized to control cells corresponding to CpG ODN A-stimulated pDCs cultures (= 100%). Expression of the cell surface markers displayed by non-activated pDCs is also represented (CpG-). (b) Levels of IFNa secretion in culture supernatants of pDCs cultured alone (control), or with HaCat, SiHa, CaSki or A431 cells were assessed by ELISA. Data represent mean percentages of relative IFNa secretion 6 SD of five independent experiments (**p < 0.01). Data were normalized to control pDCs (= 100%). (c) Relative TLR9 mRNA abundance in pDCs cultured alone (control), or with skin keratinocytes (HaCat) or SCC cell lines (SiHa, CaSki, A431). Data represent 5 independent experiments (**p < 0.01). (d) pDCs cocultured with SiHa, CaSki or A431 cells induce Treg cells differentiation in vitro. Quantitative real-time PCR analyses were performed to assess FoxP3 expression in samples collected from Treg cells induction assays. Data represent 5 independent experiments (***p < 0.001).

with R848 did not show any alterations in their maturation status and function (Supporting Information Fig. 2). Interestingly, we found that pDCs stimulated with CpG ODN A after coculture with cervical/vulvar SCC cell lines present an altered phenotype determined by monitoring expression of the surface antigens CD86, HLA-DR, CD83, CD40, CCR7 and CD11c. Indeed, we observed that, in terms of percentage of expression and mean fluorescence intensity (MFI), all pDCs maturation markers were significantly reduced after exposure to SCC cell lines when compared to control pDCs (Fig. 2a, Supporting Information Figs. 3 and 4). When compared to SCC cell lines, HaCat cells, a cell line issued from normal skin keratinocytes, slightly affected pDCs phenotype and did not differ significantly from the control. Cervical/vulvar SCC cell lines also decreased IFNa production in CpG ODN-stimulated pDCs. Indeed, pDCs cocultured with the transformed keratinocytes SiHa, CaSki and A431 showed a reduced ability to produce IFNa in response

to CpG ODN than control-pDCs (p < 0.01) (Fig. 2b). The reduced production of IFNa when exposed to SCC cell lines secretions was associated with decreased levels of TLR9 mRNA (p < 0.01) (Fig. 2c). By using a Treg cell induction assay, we demonstrated that pDCs cocultured with SCC cell lines acquire the ability to induce the differentiation of naive CD41 T cells in Treg cells. Indeed, we observed an increased expression of FoxP3 mRNA by CD41 T cells exposed to pDCs previously cocultured with SCC cell lines (Fig. 2d). Flow cytometry analyses show that FoxP3 expression on T cells is also increased at the protein level when pDCs were previously exposed to the genital cancer microenvironment (Supporting Information Fig. 5). pDCs cocultured with HaCat cells did not stimulate Treg cell differentiation from CD41 T cells. Those results are in accordance with the acquisition of a tolerogenic functionality by pDCs exposed to the cervical/vulvar tumor microenvironment. C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

351

Tumor Immunology

Demoulin et al.

Figure 3. FoxP31 Treg cells infiltrate EpM and (pre)neoplastic cervical lesions and colocalize with pDCs. (a) Analysis of Treg cells infiltration in cervical biopsy specimens by immunostaining. An anti-FoxP3 antibody was used. Magnification: 3200. (b) Semiquantitative evaluation of FoxP31 Treg cells density in the exocervical epithelium (EXO) (n 5 25), epithelial metaplasia (EpM) (n 5 16), low-grade squamous intraepithelial lesions (LSIL) (n 5 14), high-grade squamous intraepithelial lesions (HSIL) (n 5 13) and squamous cell carcinoma (SCC) (n 5 10) or in associated stromal tissues. Data represent mean numbers of FoxP31 Treg cells 6 SD in five high power (3400) fields (**p < 0.01; ***p < 0.001). (c) pDCs and Treg cells colocalize in situ in (pre)neoplastic cervical lesions. Double immunostaining using FoxP3 and BDCA-2 antibody was performed on cervical tissue specimens. FoxP31 Treg cells were stained in brown whereas BDCA-21 pDCs are visualized using permanent red. Colocalisations between BDCA-21 cells and FoxP31 cells were frequently observed in the peritumoral area. Magnification: 3200.

Treg cells infiltrate the cervical “metaplasia-dysplasiacancer” sequence and colocalize with pDCs

As pDCs exposed to a cervical/vulvar tumor microenvironment induce FoxP3 mRNA expression, we analyzed the presence of Treg cells in cervical (pre)neoplastic lesions. We performed IHC with FoxP3 antibody and showed that Treg C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

cell number increases during the cervical “metaplasiadysplasia-cancer” sequence (Fig. 3a). In the epithelial compartment, Treg cell density was higher in EpM (p < 0.001), LSIL (p < 0.01), HSIL (p < 0.01) and SCC (p < 0.001) than in the normal exocervix. In the stromal tissue, an increased number of FoxP31 Treg cells was also observed in EpM

352


(p < 0.001), LSIL (p < 0.001), HSIL (p < 0.001) and SCC (p < 0.001) compared to the normal exocervix (Fig. 3b). Double immunostaining with BDCA-2 and FoxP3 revealed the presence of a direct contact between Treg cells and pDCs, as these cells formed several clusters notably in HSIL and SCC cervical lesions (Fig. 3c).

Tumor Immunology

HMGB1 expression increases during the cervical “metaplasia-dysplasia-cancer” sequence

HMGB1 expression was investigated using IHC on cervical specimens (Fig. 4a). HSIL and SCC showed a higher expression of HMGB1 compared to the normal exocervix (p < 0.001) (Fig. 4b). Immunofluorescence experiments with cervical SCC biopsies further showed that HMGB1 is expressed in both the nucleus and cytoplasm of keratinocytes (Fig. 4c). This suggests that HMGB1 could be released in the extracellular microenvironment in order to act on pDCs. ELISA analysis on SiHa, CaSki and A431 culture media confirmed that these cell lines secrete HMGB1 (Fig. 4d). HaCat cells secreted less HMGB1 than SiHa, CaSki or A431 cells, which could explain why HaCat cells have only a little effect on pDCs phenotype and functionality in coculture experiments. To ascertain that the detection of HMGB1 is associated to its active secretion and not to cell necrosis, an annexin-PI staining was performed. As shown in Supporting Information Figure 6, a low percentage of necrotic cells was reported in our cell cultures. Next, we showed that a high percentage of pDCs generated in vitro (76.67% 6 6.04%) could respond to HMGB1 as they express the receptor for this molecule (RAGE). We analyzed pDCs expression of this receptor (% and MFI) among six different cultures of pDCs and observed an important inter-variability of RAGE expression (Fig. 4e). A similar inter-variability was also observed on pDCs isolated from the peripheral blood (data not shown). HMGB1 alters pDCs phenotype and function, an effect reversed by an anti-HMGB1 treatment

To determine whether HMGB1 could inhibit pDCs maturation and modify their functionality, pDCs were incubated in the presence of human recombinant HMGB1 (0.05; 0.25 or 2 mg ml21) and then stimulated with CpG ODN. Similar to pDCs cocultured with cervical/vulvar SCC cell lines, pDCs incubated in the presence of HMGB1 exhibited an altered phenotype characterized by a decreased expression of maturation markers such as CD86, HLA-DR, CD83, CD40, CCR7 and CD11c (Fig. 5a and Supporting Information Fig. 7). The addition of anti-HMGB1 inhibitors (DPH1.1 or A box peptide) or a blocking antibody directed against RAGE to culture assays reversed the effect of HMGB1 on pDCs and restored their phenotype. The effect of HMGB1 inhibition by quercetin was also analyzed, but a considerable cytotoxicity on pDCs (Supporting Information Fig. 8) was observed, preventing the use of this molecule in our culture systems.

We also analyzed ICOSL expression on pDCs, a molecule known to modulate immunity towards a regulatory profile. We showed that, after exposure to HMGB1 (0.25 and 2 mg ml21), pDCs increase their expression of ICOSL, corroborating our T cell induction assay results (Fig. 5a). We demonstrated that the addition of 0.25 or 2 mg ml21 of HMGB1 to the culture medium reduced significantly the expression of IFNa by pDCs, an effect reversed by the addition of an HMGB1 or RAGE inhibitor to the culture (Fig. 5b). In accordance with our coculture assay results, we showed that pDCs cultured with HMGB1 exhibit a decreased expression of TLR9 mRNA. Addition of DPH1.1 to the culture system increased TLR9 mRNA in pDCs, although without completely restoring its expression. Similar results were obtained with the addition of the A box peptide or an antiRAGE Ab to the culture (Fig. 5c). Inhibition of HMGB1 expression in cocultures restores pDCs phenotype and inhibits Treg cells induction by pDCs

To further assess the contribution of HMGB1 to pDCs functional impairment in the presence of SCC cell lines of the genital tract, cocultures were treated with DPH1.1. Figure 6a illustrates that HMGB1 inhibition reverses the effects of cervical/vulvar SCC cell lines on CD86, HLA-DR, CD83, CD40, CCR7 and CD11c maturation markers (Fig. 6a and Supporting Information Fig. 9). Similar results were obtained when the A box peptide or an anti-RAGE Ab were added to the coculture systems (Supporting Information Fig. 10). IFNa secretion and TLR9 mRNA expression were also restored when pDCs were incubated with DPH1.1 during the coculture assays (Figs. 6b and 6c). After performing Treg cell induction assays on pDCs cocultured with SCC cells lines in the presence or not of DPH1.1, we showed that HMGB1 inhibition reduced the tolerogenic activity of pDCs. Indeed, a lower induction of Treg cell differentiation from CD41 naive T cells was observed when the anti-HMGB1 antibody was added to the cocultures. These data were reported at both the mRNA (Fig. 6d) and the protein levels (Supporting Information Fig. 5). We also showed that a reduced expression of ICOSL on pDCs was associated to the inhibition of HMGB1 in the coculture assays (Fig. 6e). This down-regulation could be linked to the lower induction of Treg cells by pDCs after coculture.

Discussion HPV infection is strongly implicated in cervical carcinogenesis. About half of vulvar/vaginal and penile cancers are also HPV positive. However, HPV infection alone is not sufficient for cancer development. Additional environmental and host factors may also contribute to HPV-related carcinogenesis. Recently, it has been proposed that cervical metaplastic and/ or inflammatory cells could create an immunosuppressive environment promoting malignant transformation.2


353

Tumor Immunology

Demoulin et al.

Figure 4. HMGB1 expression is increased in HSIL and SCC when compared to the normal exocervix. (a) HMGB1 expression in paraffin-embedded sections of human cervical tissue specimens. Magnification: 3100. (b) Semiquantitative evaluation of HMGB1 expression in normal exocervical epithelium (EXO) (n 5 18), epithelial metaplasia (EpM) (n 5 13), low-grade squamous intraepithelial lesions (LSIL) (n 5 12), high-grade squamous intraepithelial lesions (HSIL) (n 5 30) and squamous cell carcinoma (SCC) (n 5 16). Asterisks indicate statistically significant differences (***p < 0.001). (c) HMGB1 expression was assessed by immunofluorescence in cervical SCC specimens. HMGB1 was detected both in the nucleus and cytoplasm of SCC cells. (d) HMGB1 is secreted in vitro by cervical/vulvar transformed keratinocytes. Secreted HMGB1 levels were assessed by ELISA in culture supernatants of human immortalized normal skin keratinocytes (HaCat) and cervical/vulvar transformed cell lines (SiHa, CaSki or A431). Data are presented as means 6 SD of three independent experiments performed in duplicate. Asterisks indicate statistically significant differences (**p < 0.01; ***p < 0.001). (e) In vitro generated pDCs express different levels of RAGE receptor. Percentages of RAGE expression on pDCs and MFI were assessed by flow cytometry on six cultures of pDCs generated from CD341 progenitor cells obtained from different donors.

pDCs play an important role in the innate immune response to viral infections and have recently been associated to tumor progression in some cancers. In the present study, C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

we investigated whether these cells play a role in the cervical cancer progression. First, we demonstrated that the density of pDCs increases in both metaplastic and (pre)neoplastic


Tumor Immunology

354

Figure 5. Human recombinant HMGB1 alters pDCs phenotype and function in culture, an effect restored by HMGB1 inhibition. (a) pDCs were incubated for 24 hr with recombinant HMGB1 (0.05, 0.25, or 2 mg ml21). In selected experiments, HMGB1 inhibitors (DPH1.1 and A box) or a neutralizing anti-RAGE antibody were added in the culture system [(DPH1.1 mAb (4 mg ml21); A box peptide (10 mg ml21); mouse monoclonal anti-RAGE Ab (10 mg ml21)]. pDCs cultured alone or in the presence of HMGB1 inhibitors or the neutralizing anti-RAGE antibody were used as controls. pDCs were then stimulated with CpG ODN A for 24 hr and expression of cell-surface molecules was measured by flow cytometric analysis. The data shown represent relative MFI of positive cells 6 SD of 10 independent experiments (*p < 0.05; **p < 0.01; ***p < 0.001). All data were normalized to control cells corresponding to stimulated-pDCs cultured alone or, for HMGB1 inhibition assays, to control cells corresponding to stimulated-pDCs cultured with the corresponding inhibitor (= 100%). (b) Levels of IFNa in culture supernatants of pDCs cultured alone, with recombinant HMGB1 (0.05, 0.25, or 2 mg ml21) and, in selected experiments, with HMGB1 inhibitors or RAGE blocking antibody were assessed by ELISA. Data represent mean percentages of relative IFNa secretion 6 SD of six independent experiments (**p < 0.01; ***p < 0.001). Data were normalized to control pDCs (= 100%). (c) Relative TLR9 mRNA abundance in pDCs cultured alone, with recombinant HMGB1 (0.05, 0.25, or 2 mg ml21) and, in selected experiments, with HMGB1 inhibitors or RAGE blocking antibody. The data shown represent six independent experiments (*p < 0.05; ***p < 0.001).

epithelial tissues as well as in the underlying stroma as compared with normal exocervix. The recruitment of pDCs observed in our biopsies could be partially explained by the increased expression of chemerin. Accordingly, chemerin

exhibited a significant chemotactic activity on pDCs at concentrations as low as 100 pM. This overexpression of chemerin observed in situ could be linked to an increased released of this molecule by local cells such as endothelial and C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

355

Tumor Immunology

Demoulin et al.

Figure 6. HMGB1 inhibition reverses the tolerogenic profile of pDCs in the cervical/vulvar tumor microenvironment. (a) Analysis of pDCs phenotype after coculture with HaCat cells or cervical/vulvar SCC cell lines treated or not with an anti-HMGB1 antibody (DPH1.1, 0.5 mg ml21). After 24 hr of coculture and 24 hr of incubation with CpG ODN, cell surface expression of CD86, HLA-DR, CD83, CD40, CCR7, CD11c was analyzed by flow cytometry. Data are from seven experiments and were normalized to pDCs cultured alone or in the presence of DPH1.1 (=100%) (*p < 0.05; **p < 0.01; ***p < 0.001). (b) Secretion of IFNa in supernatants of pDCs cocultured with HaCat or SCC cell lines in the presence or not of DPH1.1. The data shown represent mean percentages of relative IFNa secretion 6 SD of five independent experiments (*p < 0.05; **p < 0.01; ***p < 0.001). Data were normalized to pDCs cultured alone or in the presence of DPH1.1 (= 100%). (c) Relative TLR9 mRNA abundance in pDCs cocultured with SCC cell lines and incubated with or without DPH1.1. The data shown represent eight independent experiments (**p < 0.01; ***p < 0.001). (d) pDCs cocultured with SCC cell lines in the presence of DPH1.1 reduce the differentiation of Treg cells from na€ıve CD41 T cells in vitro. Quantitative real-time PCR analyses were performed to assess FoxP3 expression in samples collected from Treg cells induction assays. The data shown represent eight independent experiments (*p < 0.05; **p < 0.01; ***p < 0.001). (e) Flow cytometry analysis of ICOSL expression on pDCs after their coculture with HaCat cells or SCC cell lines in the presence or not of DPH1.1. Data were normalized to pDCs cultured alone or with DPH1.1 (= 100%) (*p < 0.05; **p < 0.01; ***p < 0.001).


Tumor Immunology

356


epithelial cells as already shown in inflamed skin.30 The signals responsible for this overexpression are still unknown but inflammatory factors secreted during the cervical “metaplasia-dysplasia-cancer” sequence31 are probably linked to this phenomenon, allowing a rapid attraction of effector cells following pathogenic signals. Proteases required for chemerin activation are likely to be present in cervical (pre)neoplastic lesions. Indeed, those lesions are often associated with an important inflammatory infiltrate that likely contains chemerin-activating proteases, such as the neutrophil serine proteases cathepsin G and elastase. Nevertheless, other chemotactic factors such as CXCL12 (SDF-1) could also be implicated in the recruitment of pDCs in cervical EpM and (pre)neoplastic lesions.32–34 Secretion of immunosuppressive molecules may be an important mechanism by which malignant cells escape from immunosurveillance. Therefore, we studied the effect of soluble factors secreted by cervical/vulvar SCC cell lines on pDCs functionality. We showed that pDCs exposed to the cervical/vulvar tumor microenvironment exhibit an impaired maturation phenotype after TLR9 but not TLR7 stimulation. Altered pDCs stimulation by TLR9 ligands only has also recently been observed in a mouse mammary tumor model.35 Although the phenotypic alterations observed in the present study look tiny, the biological consequences could be important. Indeed, accumulating data demonstrated that DC displaying a 5–20% reduction for several maturation markers induces both antigenspecific energy and regulatory properties in memory CD41 T cells.36,37 These results are in agreement with our observations. Moreover, recent studies demonstrated that sorted pDCs from both breast and ovarian tumor specimens displayed a semimature phenotype.11,38 However, several limitations (number and size of biopsies) prevent us to directly confirm our in vitro/in situ results in fresh cervical (pre)neoplastic lesions. pDCs represent the principal source of IFNa, a cytokine with one of the most potent antitumor activities. The antitumor effects of IFNa appear to be due to a combination of direct inhibition of tumor growth, differentiation and vascularization, as well as indirect immunity mediated effects.39–41 pDCs may limit the tumor spreading by IFNa secretion and suppression of this secretion represents a very effective mechanism to evade the immune response in the tumor context. As a matter of fact, several reports showed that IFNa secretion is altered or absent in some cancers.11,15,34,38 Furthermore, a selective suppression of IFNa production by pDCs in breast cancers was reported to endow pDCs with the capacity to expand FoxP31 Treg cells, contributing to tumor immune tolerance.11 As shown in the present study, SCC cell lines affect pDCs secretion of IFNa in the same way. Tumor downregulation of TLR9 expression may contribute to the reduced ability of pDCs to produce IFNa in response to CpG ODN as already demonstrated for tumor-infiltrating pDCs in head and neck cancers.34 Our IHC analysis showed a progressive increase in the number of Treg cells according to the severity of cervical

(pre)neoplastic lesions. Furthermore, we observed that pDCs and Treg cells colocalize in both metaplastic and (pre)neoplastic lesions of the cervix. In line with these observations, we observed that pDCs exposed to the cervical/vulvar tumor microenvironment promote the differentiation of Treg cells from na€ıve CD41 T cells as shown by the increased FoxP3 expression (mRNA and protein levels). One explanation for this colocalization could be that these cell clusters around (pre)neoplastic keratinocytes reflect the creation of local tertiary lymph node tissues leading to the generation of Treg cells, as already shown in renal cell carcinoma.42 Several studies have implicated ICOSL in the differentiation or activation of Treg cells by pDCs.43 Here, we showed that pDCs cocultured with cervical/vulvar SCC cell lines display an increased expression of this molecule, when compared to control pDCs, which could explain local Treg cells expansion from CD41 T cells during cervical cancer progression. In addition to the local expansion of Treg cells, recruitment of these cells from lymph nodes through the expression of CXCL12 (SDF-1), already shown in cervical (pre)neoplastic lesions,44 might also increase Treg cell accumulation during this cancer process. The intratumoral production of immunosuppressive molecules may be an important mechanism by which cancer cells escape from the immunosurveillance. Among the factors potentially responsible for pDCs alterations is PGE2, notably produced in cervical (pre)neoplastic lesions and by several cervical cell lines.45 PGE2 may promote the carcinogenesis by different mechanisms, such as apoptosis inhibition, cell proliferation, angiogenesis and, more interestingly, immunosuppression.46 Bekeredjian-Ding et al. have reported that pDCs treated with oral SCC supernatants that contain high levels of PGE2 and TGFb present a modulation of their phenotype and an inhibition of IFNa secretion.47 Addition of recombinant PGE2 to our pDCs cultures did not mimic the effect of cervical/vulvar SCC cell lines secretions as shown by the slight modulation of pDCs phenotype, the weak effect on IFNa secretion and the induction of TLR9 mRNA upregulation in pDCs (Supporting Information Fig. 11). Differences between our results and data from the literature may be explained by the origin of pDCs used. Indeed, pDCs used in our culture assays are generated from CD341 hematopoietic progenitor cells isolated from cord blood. In contrast, experiments from Bekeredjian-Ding et al. study were performed with pDCs from the peripheral blood. A differential response to stimuli has already been observed between dendritic cells (DC) generated from CD341 cells and DC derived from peripheral blood monocytes.48 Among other immunosuppressive factors potentially responsible for pDCs alterations in the cervical/vulvar tumor microenvironment, HMGB1 is an interesting candidate. Indeed, the serum level of this molecule is not only inversely correlated with disease-free and overall survival in women presenting SCC of the uterine cervix,24 but HMGB1 was also reported to suppress pDCs cytokine secretion and maturation in response to TLR9 agonists.23 After having confirmed that C 2014 UICC Int. J. Cancer: 137, 345–358 (2015) V

HMGB1 is secreted by the different cell lines (SiHa, CaSki, A431) used in our coculture assays, we demonstrated, for the first time, a significant in situ increase of HMGB1 expression during cervical (pre)cancer progression. Interestingly, an overexpression of HMGB1 was also reported in a large proportion of cancer types (for review, see Ref. [49]). We confirmed, by using a recombinant molecule, that HMGB1 alters both pDCs phenotype and function and that the inhibition of its action in our cocultures restores the proper functionality of pDCs. Indeed, HMGB1 modified pDCs functionality and rendered them tolerogenic. As recently shown, this molecule might also directly promote activity of Treg cells.50 Those two phenomena might act in synergy and exacerbate Treg cells induction in the cervical/vulvar cancer microenvironment. As shown by the use of a blocking antibody, HMGB1 acts on pDCs through RAGE receptor. Interestingly, we showed that RAGE expression present an inter-variability between pDCs cultures from different donors which could explain the variations (characterized by the high standard deviations) observed in studies focusing on pDCs functionality. Interestingly, this inter-variability was also found in pDCs isolated from the peripheral blood (data not shown). HMGB1 is known to present a nuclear localization in normal conditions and its affinity for DNA would be regulated by phosphorylation and acetylation. The presence of HMGB1 in the extracellular milieu during cancer progression could be due to passive diffusion from necrotic cells in hypoxia but also to an active transport from both immune (e.g., macrophages and DC) and cancer cells. It has been proposed that active secretion of HMGB1 from immune cells is mediated by unconventional vesicular secretion.19 Mechanisms of cytoplasmic HMGB1 transport and extracellular secretion by cancer cells are poorly defined but phosphorylation of HMGB1 was proposed as a potential step in this phenomenon.51 We further determined if HMGB1 expression is linked to carcinogenesis or to HPV infection. After having shown by

ELISA that both HPV1 (SiHa, CaSki) and HPV2 cell lines (A431) express HMGB1, we analyzed its expression by immunohistochemistry in high-grade vulvar intraepithelial neoplasia (VINII and VINIII) and demonstrated that HMGB1 expression is similar in HPV1 (p161) and in HPV2 (p162) vulvar lesions (Supporting Information Fig. 12). These results suggest that HMGB1 is not related to HPV viral oncoprotein expression. In conclusion, it appears that pDCs are recruited during the early steps of cervical transformation and probably play a role in the antiviral response against HPV, favoring the clearance of the preneoplastic lesion. Indeed, it has been shown that pDCs produce IFNa in response to HPV16 virus-like particles.52 Nevertheless, secretion of HMGB1 during the cancer progression potentially prevents their antiviral response through both the inhibition of IFNa secretion and the increased expression of ICOSL. pDCs expressing ICOSL may promote accumulation of Treg cells in the tumor setting and lead to the development of a tolerogenic microenvironment. Indeed, as tumors have been shown to express danger signals that may activate pDCs such as LL37, endogenous selfnucleic acids released from dying tumor cells or fragmented mitochondrial DNA released from apoptotic cells,53,54 we can speculate that upon complexes recognition, pDCs also exposed to HMGB1 may be improperly activated and exacerbate the process already in place. Treatments targeting HMGB1 (neutralizing anti-HMGB1 antibody or glycyrrhizin, a naturally specific inhibitor of extracellular HMGB1) could be used to restore antitumor activities of pDCs and, therefore, overcome immune tolerance in cancer microenvironment.

Acknowledgement The authors thank the University of Lie`ge Biobank, the GIGA-imaging and flow cytometry facility, and Estelle Dortu from the GIGA-immunohistology platform for her excellent technical assistance.

References 1.

Ellerbrock TV, Chiasson MA, Bush TJ, et al. Incidence of cervical squamous intraepithelial lesions in HIV-infected women. JAMA 2000;283:1031–7.

2.

Herfs M, Hubert P, Moutschen M, et al. Mucosal junctions: open doors to HPV and HIV infections? Trends Microbiol 2011;19:114–20.

3.

Herfs M, Hubert P, Kholod N, et al. Transforming growth factor-beta1-mediated Slug and Snail transcription factor up-regulation reduces the density of Langerhans cells in epithelial metaplasia by affecting E-cadherin expression. Am J Pathol 2008;172:1391–402.

4.

Herman L, Hubert P, Caberg JH, et al. MIP3 alpha stimulates the migration of Langerhans cells in models of human papillomavirus (HPV)associated (pre)neoplastic epithelium. Cancer Immunol Immunother 2007;56:1087–96.

5.

Colonna M, Trinchieri G, Liu YJ. Plasmacytoid dendritic cells in immunity. Nat Immunol 2004;5: 1219–26.

6.

Demoulin S, Herfs M, Delvenne P, et al. Tumor microenvironment converts plasmacytoid dendritic cells into immunosuppressive/tolerogenic cells: insight into the molecular mechanisms. J Leukoc Biol 2013;93:343–52. 7. von Marschall Z, Scholz A, Cramer T, et al. Effects of interferon alpha on vascular endothelial growth factor gene transcription and tumor angiogenesis. J Natl Cancer Inst 2003;95:437–48. 8. Shao X, Liu C. Influence of IFN-alpha and IFNgamma on lymphangiogenesis. J Interferon Cytokine Res 2006;26:568–74. 9. U’Ren L, Guth A, Kamstock D, et al. Type I interferons inhibit the generation of tumorassociated macrophages. Cancer Immunol Immunother 2010;59:587–98. 10. Liang S, Wei H, Sun R, et al. IFNalpha regulates NK cell cytotoxicity through STAT1 pathway. Cytokine 2003;23:190–9. 11. Sisirak V, Faget J, Gobert M, et al. Impaired IFNa production by plasmacytoid dendritic cells


favors regulatory T cell expansion and contributes to breast cancer progression. Cancer Res 2012;72: 5188–97. 12. Curiel TJ, Cheng P, Mottram P, et al. Dendritic cell subsets differentially regulate angiogenesis in human ovarian cancer. Cancer Res 2004;64:5535– 8. 13. Sawant A, Hensel JA, Chanda D, et al. Depletion of plasmacytoid dendritic cells inhibits tumor growth and prevents bone metastasis of breast cancer cells. J Immunol 2012;189:4258–65. 14. Jensen TO, Schmidt H, Moller HJ, et al. Intratumoral neutrophils and plasmacytoid dendritic cells indicate poor prognosis and are associated with pSTAT3 expression in AJCC stage I/II melanoma. Cancer 2011;118:2476–85. 15. Treilleux I, Blay JY, Bendriss-Vermare N, et al. Dendritic cell infiltration and prognosis of early stage breast cancer. Clin Cancer Res 2004;10: 7466–74.

Tumor Immunology

357

Demoulin et al.

Tumor Immunology

358

16. Labidi-Galy SI, Treilleux I, Goddard-Leon S, et al. Plasmacytoid dendritic cells infiltrating ovarian cancer are associated with poor prognosis. Oncoimmunology 2012;1:380–2. 17. Scaffidi P, Misteli T, Bianchi ME. Release of chromatin protein HMGB1 by necrotic cells triggers inflammation. Nature 2002;418:191–5. 18. Dumitriu IE, Baruah P, Valentinis B, et al. Release of high mobility group box 1 by dendritic cells controls T cell activation via the receptor for advanced glycation end products. J Immunol 2005;174:7506–15. 19. Gardella S, Andrei C, Ferrera D, et al. The nuclear protein HMGB1 is secreted by monocytes via a non-classical, vesicle-mediated secretory pathway. EMBO Rep 2002;3:995–1001. 20. Kang R, Zhang Q, Zeh HJ, III, et al. HMGB1 in cancer: good, bad, or both? Clin Cancer Res 2013; 19:4046–57. 21. Wild CA, Bergmann C, Fritz G, et al. HMGB1 conveys immunosuppressive characteristics on regulatory and conventional T cells. Int Immunol 2012;24:485–94. 22. Kusume A, Sasahira T, Luo Y, et al. Suppression of dendritic cells by HMGB1 is associated with lymph node metastasis of human colon cancer. Pathobiology 2009;76:155–62. 23. Popovic PJ, DeMarco R, Lotze MT, et al. High mobility group B1 protein suppresses the human plasmacytoid dendritic cell response to TLR9 agonists. J Immunol 2006;177:8701–7. 24. Sheng X, Du X, Zhang X, et al. Clinical value of serum HMGB1 levels in early detection of recurrent squamous cell carcinoma of uterine cervix: comparison with serum SCCA, CYFRA21-1, and CEA levels. Croat Med J 2009;50:455–64. 25. Skrzeczynska-Moncznik J, Wawro K, Stefanska A, et al. Potential role of chemerin in recruitment of plasmacytoid dendritic cells to diseased skin. Biochem Biophys Res Commun 2009;380:323–7. 26. Detry C, Waltregny D, Quatresooz P, et al. Detection of bone sialoprotein in human (pre)neoplastic lesions of the uterine cervix. Calcif Tissue Int 2003;73:9–14. 27. Demoulin S, Roncarati P, Delvenne P, et al. Production of large numbers of plasmacytoid dendritic cells with functional activities from CD34(1) hematopoietic progenitor cells: use of interleukin3. Exp Hematol 2012;40:268–78. 28. Guo X, Jie Y, Ren D, et al. In vitro-expanded CD4(1)CD25(high)Foxp3(1) regulatory T cells controls corneal allograft rejection. Hum Immunol 2012;73:1061–7. 29. Hippen KL, Merkel SC, Schirm DK, et al. Generation and large-scale expansion of human induci-


30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

ble regulatory T cells that suppress graft-versushost disease. Am J Transplant 2011;11:1148–57. Vermi W, Riboldi E, Wittamer V, et al. Role of ChemR23 in directing the migration of myeloid and plasmacytoid dendritic cells to lymphoid organs and inflamed skin. J Exp Med 2005;201: 509–15. Herfs M, Hubert P, Delvenne P. Epithelial metaplasia: adult stem cell reprogramming and (pre)neoplastic transformation mediated by inflammation? Trends Mol Med 2009;15:245–53. Zou W, Machelon V, Coulomb-L’Hermin A, et al. Stromal-derived factor-1 in human tumors recruits and alters the function of plasmacytoid precursor dendritic cells. Nat Med 2001;7:1339– 46. Vermi W, Bonecchi R, Facchetti F, et al. Recruitment of immature plasmacytoid dendritic cells (plasmacytoid monocytes) and myeloid dendritic cells in primary cutaneous melanomas. J Pathol 2003;200:255–68. Hartmann E, Wollenberg B, Rothenfusser S, et al. Identification and functional analysis of tumorinfiltrating plasmacytoid dendritic cells in head and neck cancer. Cancer Res 2003;63:6478–87. Le Mercier I, Poujol D, Sanlaville A, et al. Tumor promotion by intratumoral plasmacytoid dendritic cells is reversed by TLR7 ligand treatment. Cancer Res 2013;73:4629–40. Watkins SK, Zhu Z, Riboldi E, et al. FOXO3 programs tumor-associated DCs to become tolerogenic in human and murine prostate cancer. J Clin Invest 2011;121:1361–72. Bellone G, Carbone A, Smirne C, et al. Cooperative induction of a tolerogenic dendritic cell phenotype by cytokines secreted by pancreatic carcinoma cells. J Immunol 2006;177:3448–60. Labidi-Galy SI, Sisirak V, Meeus P, et al. Quantitative and functional alterations of plasmacytoid dendritic cells contribute to immune tolerance in ovarian cancer. Cancer Res 2011;71:5423–34. Kirkwood J. Cancer immunotherapy: the interferon-alpha experience. Semin Oncol 2002;29: 18–26. Ferrantini M, Capone I, Belardelli F. Interferonalpha and cancer: mechanisms of action and new perspectives of clinical use. Biochimie 2007;89: 884–93. Dunn GP, Bruce AT, Sheehan KC, et al. A critical function for type I interferons in cancer immunoediting. Nat Immunol 2005;6:722–9. Middel P, Brauneck S, Meyer W, et al. Chemokine-mediated distribution of dendritic cell subsets in renal cell carcinoma. BMC Cancer 2010; 10:578.

43. Faget J, Bendriss-Vermare N, Gobert M, et al. ICOS-ligand expression on plasmacytoid dendritic cells supports breast cancer progression by promoting the accumulation of immunosuppressive CD41 T cells. Cancer Res 2012;72:6130–41. 44. Jaafar F, Righi E, Lindstrom V, et al. Correlation of CXCL12 expression and FoxP31 cell infiltration with human papillomavirus infection and clinicopathological progression of cervical cancer. Am J Pathol 2009;175:1525–35. 45. Herfs M, Herman L, Hubert P, et al. High expression of PGE2 enzymatic pathways in cervical (pre)neoplastic lesions and functional consequences for antigen-presenting cells. Cancer Immunol Immunother 2009;58:603–14. 46. Nakanishi M, Rosenberg DW. Multifaceted roles of PGE2 in inflammation and cancer. Semin Immunopathol 2013;35:123–37. 47. Bekeredjian-Ding I, Schafer M, Hartmann E, et al. Tumour-derived prostaglandin E and transforming growth factor-beta synergize to inhibit plasmacytoid dendritic cell-derived interferonalpha. Immunology 2009;128:439–50. 48. Power CA, Church DJ, Meyer A, et al. Cloning and characterization of a specific receptor for the novel CC chemokine MIP-3alpha from lung dendritic cells. J Exp Med 1997;186:825–35. 49. Tang D, Kang R, Zeh HJ, III, et al. High-mobility group box 1 and cancer. Biochim Biophys Acta 2010;1799:131–40. 50. Wild CA, Brandau S, Lotfi R, et al. HMGB1 is overexpressed in tumor cells and promotes activity of regulatory T cells in patients with head and neck cancer. Oral Oncol 2012. 51. Kang HJ, Lee H, Choi HJ, et al. Non-histone nuclear factor HMGB1 is phosphorylated and secreted in colon cancers. Lab Invest 2009;89: 948–59. 52. Lenz P, Lowy DR, Schiller JT. Papillomavirus virus-like particles induce cytokines characteristic of innate immune responses in plasmacytoid dendritic cells. Eur J Immunol 2005;35:1548–56. 53. Lande R, Gregorio J, Facchinetti V, et al. Plasmacytoid dendritic cells sense self-DNA coupled with antimicrobial peptide. Nature 2007;449:564– 9. 54. Ries M, Schuster P, Thomann S, et al. Identification of novel oligonucleotides from mitochondrial DNA that spontaneously induce plasmacytoid dendritic cell activation. J Leukoc Biol 2013;94: 123–35.
