Cervical Cancer–Instructed Stromal Fibroblasts Enhance IL23 Expression in Dendritic Cells to Support Expansion of Th17 Cells

Cervical carcinogenesis is a consequence of persistent infection with genital high-risk human papillomavirus (HPV). Cervical cancer develops through cervical intraepithelial neoplasia (CIN1–3) in a well-characterized multistep process that takes years or decades.

As a mechanism of immune escape, the high-risk HPV oncoproteins interfere with inflammatory signaling pathways in keratinocytes. They suppress the expression of chemokines, including monocyte chemoattractant protein-1 (CCL2) and macrophage inflammatory protein-3α (CCL20), eventually preventing the recruitment of antigen-presenting cells to the epithelium (1–3). Thus, inflammatory responses are rarely observed in persisting low-grade lesions.

At later stages of cervical carcinogenesis, precancerous high-grade lesions and invasive cervical cancers are often associated with strong inflammatory infiltrates in the stroma (4–10). Different studies have shown that HPV-transformed keratinocytes actively contribute to the inflammatory microenvironment during cervical carcinogenesis via production of the cytokine IL6 (6, 8, 9, 11). IL6 is expressed in high-grade as well as invasive lesions (12). It preferentially acts in a paracrine manner at later stages of the disease (12) to suppress the immune functions in myeloid immune cells, such as CCR7-dependent migration of dendritic cells (DC), but enhances their tumor-promoting activities, such as MMP-9 production (6–8). Furthermore, we recently demonstrated that cervical cancer cells actively contribute to the recruitment of Th17 cells. Although HPV-infected keratinocytes express only low amounts of the Th17-attracting chemokine CCL20 (2, 9), they instruct cervical fibroblasts to produce CCL20 in an IL6/CAAT/enhancer-binding protein β (C/EBPβ)–dependent manner and thereby, support Th17 cell recruitment (9). Cervical cancer–infiltrating Th17 cells in situ were correlated with advanced International Federation of Gynecology and Obstetrics (FIGO; ref. 13) stages of the tumors (9) and a poor prognosis for the patients (14).

Th17 cells were identified as a subgroup of Th cells (15) exhibiting proinflammatory as well as tumor-promoting properties in different cancer types (16, 17). Human Th17 cell differentiation from naïve T cells is dependent on the cytokines IL6, TGFβ, IL1β, and IL23 (18, 19). IL23 has been described to be necessary for the expansion, activity and cytokine release of Th17 cells (20). IL23 is a member of the IL12 family, which has the ability to modulate T-cell activity (21). It is a heterodimeric cytokine mainly expressed by DCs containing the subunit IL23p19 and shares the subunit IL12p40 with IL12, a cytokine crucial for Th1 polarization (20). Similar to IL17, serum levels of IL23 were shown to be increased in patients with cervical cancer (22).

Previously, we detected CD83-positive DCs in cervical cancer biopsies and demonstrated that cervical cancer cells suppress their CCR7 expression (8), thus promoting the accumulation of phenotypically mature but functionally impaired DCs in the tumor stroma (8). In the peripheral tissue, DCs are in permanent contact with their tissue microenvironment interacting with stromal cells, that is, fibroblasts. Fibroblasts can influence DC functions and their cytokine and chemokine secretion patterns (23–25).

In this study, we investigated the mechanism underlying the expansion of Th17 cells in cervical cancers. We demonstrate that the presence of CD83⁺IL23p19⁺ DCs in cervical cancers in situ correlates with the presence of tumor-infiltrating Th17 cells. Our data provide evidence that cervical cancer cells differentially regulate the expression of the T-cell polarizing cytokines IL23 and IL12 in DCs. Although cervical carcinoma cell–derived IL6 suppressed the IL12 expression by inhibition of the IL12p35 subunit in DCs, it induced the expression of C/EBPβ-mediated IL1β production of fibroblasts. As a consequence, fibroblasts produced IL1β-induced expression of the Th17-expanding cytokine IL23 by cervical cancer–instructed DCs. This led to the expansion of memory CD4⁺IL17⁺ T cells in an IL23-dependent manner. To our knowledge, this is the first report on a mechanism how cervical cancer cells can support the expansion of Th17 cells.

This study has been conducted according to Declaration of Helsinki principles. IHC and immunofluorescence stainings of anonymized tissue samples and the usage of primary human cervical fibroblasts or peripheral blood mononuclear cells (PBMC) were approved by the Ethics Committees of the Medical Faculty of the Saarland University at the Saarland Ärztekammer (Saarbrücken, Germany). Written informed consent was provided by all study participants.

Formalin-fixed paraffin-embedded anonymized lesions of the cervix uteri from 35 patients were taken from the local pathology archive of the Saarland University Medical Center (Homburg, Germany). Histologic classification was ascertained by expert pathologists (Y.-J. Kim or R.M. Bohle). Lesions were stained with pancytokeratin antibody and costained for CD4 and IL17, CD83, and IL23p19 or α-smooth muscle actin (α SMA) and IL1β or CD83 as described in Supplementary Materials and Methods. To evaluate the number of infiltrating Th17 cells or CD83⁺IL23p19⁺cells, five randomized pictures (200×) were taken per biopsy and the number of CD4⁺IL17⁺ or CD83⁺IL23p19⁺ cells was counted.

HPV18-positive cervical carcinoma cell lines SW756 (ATCC CRL-10302), HeLa (ATCC CCL-2), and HPV16-positive SiHa (ATCC HTB-35) and CaSki (ATCC CRL-1550) were obtained from M. von Knebel-Doeberitz (Department of Applied Tumour Biology, Institute of Pathology, University Hospital, Heidelberg, Germany) before 2000 were authenticated by qRT-PCR for HPV16 or HPV18 E6 and E7 expression and by the German collection of microorganisms and cell cultures (DSMZ) using short tandem repeat DNA typing in May 2017. Normal exocervical keratinocytes (NECK), normal exocervical fibroblasts (NECF), and fibroblasts from cervical cancer outgrowth cultures were cultured as previously described (9, 26, 27). Cell lines were passaged for less than 3 months. NECFs were used in passage 2 after isolation. Fibroblasts from cervical cancer outgrowth cultures were directly used without passaging. The isolated fibroblasts were characterized by qPCR and immunofluorescence by evaluation of stromal, epithelial, endothelial, and immune cell markers (Supplementary Materials and Methods; Supplementary Figs. S1A–S1D and S2A and S2B). For conditioned media, cells were cultured at a density of 1 × 10⁶/mL. After 24 hours, fresh RPMI1640 medium (Sigma) plus supplements [10% heat-inactivated endotoxin-tested FCS (Biochrom) and 1 mmol/L sodium pyruvate] was added. Conditioned media were collected 24 hours later.

NECFs were stimulated for 24 hours with conditioned media of cervical cancer cells or NECK or plain medium as a control. Cells were washed and cultured again in fresh RPMI. Twenty-four hours later, supernatants were collected. Neutralizing anti-IL6 antibody or matched isotype control antibody (10 μg/mL; R&D Systems) was added to conditioned media 30 minutes or 2 hours before usage as indicated. A total of 10 pmol of indicated siRNAs (ON-TARGETplus Non-targeting siRNA #2 and ON-TARGET smart-pool for C/EBPβ, all from Thermo Fisher Scientific) was transfected with Lipofectamine RNAiMax (Invitrogen). Twenty-four hours posttransfection, cells were stimulated as described above.

Monocytes were isolated out of whole blood from healthy donors as described previously (8). To generate cervical cancer–instructed DCs (8) or fibroblast-instructed DCs, monocytes were cultured at a density of 1 × 10⁶ cells/mL for 6 days in six‐well plates in RPMI medium plus supplements, GM-CSF (100 ng/mL Leukine; Berlex), and IL4 (5 ng/mL; Miltenyi Biotec). Conditioned media from cervical cancer cells or from fibroblasts from cervical cancer outgrowth cultures (final dilution of 50% vol/vol) or recombinant human IL6 (100 ng/mL; PeproTech) were added from the first day of culture. Cells were fed every 2 days with medium, conditioned media, or rhIL6. In neutralization experiments, neutralizing anti-human IL6 monoclonal or isotype control antibody (10 μg/mL; R&D Systems) was added to conditioned media 30 minutes before usage. DC maturation was induced with TNFα (500 U/mL; Bender & Co.) on day 6 for 8 hours, followed by three washing steps and used for coculture experiments.

For coculture experiments, NECFs were seeded at a density of 1 × 10⁶ cells/mL. In same experiments, NECFs were prestimulated with conditioned media of cervical cancer cells. After 24 hours, 1 × 10⁶ cells/mL normal or cervical cancer–instructed immature DCs (iDC) or mature DCs (mDC) were added. As a control, DCs and fibroblasts were cultured alone under the same conditions. After 16 hours, supernatants were harvested and cells were collected for flow cytometry analysis or RNA isolation. For cell separation, DCs were removed from adherent fibroblast monolayers. DCs and fibroblasts were washed three times with PBS. Purity of DCs was determined by flow cytometry using the DC-specific marker CD1a and was 86.7% ± 7.2% in averaged three independent experiments. To control purity of fibroblasts and to exclude contamination with DCs, fibroblasts were stained with CD1a. Fibroblasts showed no expression of the DC-specific marker. In neutralizing experiments, neutralizing anti-human IL1β monoclonal or isotype control antibody (2 μg/mL; R&D Systems) was added to supernatants of prestimulated NECFs 2 hours before coculture.

RNA isolation, cDNA synthesis, real-time PCR, and normalization to β-actin or RPL13A were performed as described previously (9, 26). The 75-bp fragment of IL12B was detected with primers 5′-CCCTGACATTCTGCGTTCA-3′ and 5′-AGGTCTTGTCCGTGAAGACTCTA-3′ and probe no. 37; the 91-bp fragment of IL12A with primers 5′-CACTCCCAAAACCTGCTGAG-3′ and 5′-CAATCTCTTCAGAAGTGCAAGG-3′ and probe no. 50; the 71-bp fragment of IL23A with primers 5′-AGCTTCATGCCTCCCTACTG-3′ and 5′-CTGCTGAGTCTCCCAGTGGT-3′, and probe no. 30; and the 70-bp fragment of IL1β with primers 5′-CTGTCCTGCGTGTTGAAAGA-3′ and 5′-TTGGGTAATTTTTGGGATCTACA-3′ and probe no. 78.

IL12p70, IL23, and IL1β concentrations were determined with DuoSet (R&D Systems) according to the manufacturer's instructions. Detection limits were 31.25 pg/mL, 125 pg/mL, or 3.91 pg/mL, respectively.

Naïve or memory CD4⁺ T cells were isolated by negative selection from fresh PBMCs using human naïve CD4⁺ T-Cell Isolation Kit II or Memory CD4 T Cell Isolation Kit (Miltenyi Biotec), respectively. Purity was determined with anti-CD4, anti-CD45RO, and anti-CD45RA and was 90.2% ± 4.2% for naïve and 92.8% ± 5.7% for memory T cells in averaged three independent experiments. Cells were cultured in RPMI1640 medium supplemented with 10% heat-inactivated endotoxin-tested FCS (Biochrom).

Supernatants from DC–NECF cocultures (final dilution of 50% vol/vol) or 100 ng/mL rhIL23 (Miltenyi Biotec) as a control were added to 1 × 10⁵ naïve or memory CD4-positive T cells/96-well from the first day of culture in the presence of beads from the T-cell Activation/Expansion Kit (Miltenyi Biotec). Cells were fed every 2 days with cocultures supernatants or rhIL23. In neutralization experiments, anti-human IL23 or isotype control antibody (1 μg/mL, R&D Systems) was added to coculture supernatants 30 minutes before usage. On day 6, T cells were restimulated with phorbol-12-myristate-13-acetate (PMA; 5 ng/mL)/ionomycin (500 ng/mL) (both from Sigma) for 6 hours. After 2 hours, brefeldin A (10 μg/mL; Sigma) was added. Supernatants were collected and analyzed by ELISA and cells were stained for flow cytometry analysis.

Stimulated T cells were treated with 2 mmol/L EDTA for 15 minutes to disrupt cell–cell interactions, fixed with 3% paraformaldehyde, and stained under permeabilizing conditions using anti-CD4-FITC, anti-IL17-APC, and anti-IFNγ-BV421 or anti-CD4-APC and anti-IFNγ-FITC or respective conjugated isotype control antibodies (BD Biosciences and Miltenyi Biotec) and analyzed by flow cytometry (FACSCantoII; BD Biosciences). Numbers of Th17 cells were determined using Trucount Absolute Counting Tubes (BD Biosciences) according to the manufacturer's suggestions.

All statistical analyses were performed using the GraphPad Prism 5 (GraphPad Software) program. To evaluate the statistical differences between the analyzed groups, a two-sided t test was applied for the comparison between two groups and the one-way ANOVA test (with Bonferroni posttest) for comparison of >2 groups. Significances are indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Correlation between the number of CD83⁺IL23p19⁺ cells, Th17 cells, and FIGO (13) stages of squamous cell carcinomas (SCC) was done using Spearman rank correlation.

Thirty-five cervical cancer biopsies were stained for CD83/IL23p19 and CD4/IL17 (Th17 cells), respectively using double immunofluorescence (Fig. 1A; isotype control stainings in Supplementary Fig. S3). A total of 76% of IL23p19⁺ cells were also positive for CD83. The absolute numbers of CD83⁺IL23p19⁺ cells significantly correlated with the presence of Th17 cells in the stroma (r = 0.5100; P = 0.0040; Fig. 1B), whereas the correlation between total numbers of CD83⁺ cells and Th17 cells was lower (r = 0.4413; P = 0.0080; Fig. 1C) and no correlation was between the numbers of CD83⁻/IL23p19⁺ cells and Th17 cells (r = −0.09848; P = 0.6787; Fig. 1D). The strongest correlation was observed when we evaluated the percentage of IL23p19-expressing CD83⁺ cells per total CD83⁺ cells and percentage of IL17-expressing CD4⁺ cells per total CD4⁺ cells (r = 0.6946; P < 0.0001; Fig. 1E). Moreover, the presence of tumor-infiltrating IL23p19-expressing CD83⁺ and Th17 cells both significantly correlated with more advanced FIGO stages of the respective biopsies (Fig. 1F and G) and patients with lymph node metastasis showed significantly more CD83⁺IL23p19⁺ or Th17 cells (Fig. 1H and I). Notably, retrospective analysis demonstrated that in patients developing recurrence of cervical cancers, IL23p19-expressing CD83⁺ cells per total CD83⁺ cells (Fig. 1J) and IL17-expressing CD4⁺ cells per total CD4⁺ cells (Fig. 1K) were significantly more frequent in their tumor tissues than in patients without relapse. Thus, the results from our in situ analysis indicated that the IL23 expression of mature CD83⁺ DCs might be linked to enhanced proportions of IL17-producing CD4⁺ T cells within total CD4⁺ cells during cervical cancer progression.

To analyze the IL23 regulation in DCs in the cervical tumor micromilieu in vitro, monocytes were cultivated for 6 days with IL4 and GM-CSF in the absence or presence of conditioned media of different HPV16-positive (SiHa and CaSki) or HPV18-positive (SW756 and HeLa) cervical cancer cells to generate normal or cancer-instructed DCs (8). To induce DC maturation, cells were stimulated with TNFα for 8 hours (Fig. 2A). Because 77.6% of CD83⁺ DCs were in contact with stromal cells in the cervical tumor micromilieu interacting with αSMA–positive myofibroblasts in our in situ analysis (Fig. 2B), we analyzed the impact of fibroblast on the IL23 expression in cocultures of mDCs and NECFs.

Normal and cervical cancer–instructed iDCs produced low levels of IL23 (up to 171 pg/mL). Stimulation with TNFα alone induced the expression of IL23 in normal and in cervical cancer–instructed DCs (up to 556 pg/mL). The addition of NECFs to DCs for 16 hours significantly increased the IL23 secretion in cocultures with normal and cervical cancer cell–instructed mDCs (up to 3,092 pg/mL; Fig. 2C, gray bars). Interestingly, when we used fibroblasts that were prestimulated with conditioned media of cervical cancer cells, the IL23 expression was further increased in mDC–fibroblast cocultures (up to 6,878 pg/mL; Fig. 2C, black bars). Notably, we found a significantly enhanced IL23 expression in cocultures of normal mDCs and fibroblasts from cervical cancer outgrowth cultures (black bars, Fig. 2D) in comparison with cocultures with NECFs (Fig. 2D, gray bars).

In contrast to IL23, the Th1-polarizing cytokine IL12 was regulated in an opposite manner. Normal and cervical cancer–instructed iDCs produced very low levels of IL12 (up to 32 pg/mL). TNFα stimulation induced the IL12 expression in normal DCs up to 500 pg/mL (white bar) and the addition of NECFs (gray bar) or prestimulated fibroblasts (black bar) significantly induced the IL12 expression in cocultures with normal mDCs (Fig. 2E). In contrast, the IL12 expression was significantly lower in cervical cancer–instructed mDCs (102–140 pg/mL; Fig. 2E, white bars) and the addition of NECFs (Fig. 2E, gray bars) or prestimulated fibroblasts (Fig. 2E, black bars) did not increase the IL12 expression in cocultures with cancer-instructed mDCs (216–396 pg/mL). A similarly differential IL23 versus IL12 regulation pattern was obtained after using the TLR4 agonist lipopolysaccharide (LPS) as a DC maturation stimulus (Supplementary Fig. S4A and S4B).

To investigate whether efficient DC maturation took place in the presence of NECFs, the expression of maturation markers was analyzed by flow cytometry. After TNFα or LPS stimulation, CD83, CD80, CD86, MHC I, and MHC II were upregulated in normal and to similar levels in cancer-instructed mDCs. Cocultivation with fibroblasts had no impact on maturation marker expression but slightly enhanced the MHC I expression (Supplementary Table S1). In summary, we found that coculture of normal mDCs with NECFs enhances IL12 and IL23 expression. Cervical cancer cells however, selectively interfere with IL12 expression and rather promote IL23 expression in mDC–fibroblast cocultures.

To study the differential regulation of IL12 and IL23, DCs and NECFs were separated after cocultivation and mRNA was prepared from both cell types. As expected, DCs were the main producers of IL12p35, IL12p40, and IL23p19 transcripts in DC–NECF cocultures but with varying expression levels (Fig. 3A–C). Expression of the IL23p19 subunit was detectable in normal and in cervical cancer–instructed mDCs after coculture with NECFs (up to 2.4-fold expression; Fig. 3A, gray bars). The expression was further increased in mDCs after coculture with fibroblasts prestimulated with conditioned media of cervical cancer cells (up to 5.8-fold; Fig. 3A, black bars). The same expression pattern in mDCs was found for the IL12p40 subunit shared by IL12 and IL23 (Fig. 3B). However, after coculture, IL12p35 production was only induced in normal mDCs. In contrast, the expression of the IL12p35 subunit was significantly reduced in cervical cancer–instructed mDCs, independently of whether they were cocultured with NECFs or with prestimulated fibroblasts (up to 97.4% reduction; Fig. 3C). Similarly, results concerning IL23p19, IL12p40, and IL12p35 expression were obtained after using the TLR4 agonist LPS as a DC maturation stimulus (Supplementary Fig. S4C–S4E). In conclusion, cocultivation of fibroblasts with mDCs increases the IL23p19 and IL12p40 expression in DCs, but cervical cancer cells interfere with IL12 production by suppression of the IL12p35 subunit.

Next, we were interested in the nature of the soluble factors present in the conditioned media of cervical cancer cells suppressing the IL12 expression in mDCs. We recently demonstrated that cervical cancer cell–derived IL6 suppresses the transcription factor NF-κB in phenotypically mDCs (8). Because NF-κB is a regulator of IL12p35 (28) and cervical cancer cells express high amounts of IL6 in vitro (9, 12), the cytokine IL6 was regarded as an interesting candidate. In mDCs, differentiated in the presence of increasing amounts of recombinant human IL6 (rhIL6; 6.2-5-100 ng/mL), the IL12 expression was dose-dependently suppressed on protein level after coculture with NECFs (up to 96%; Fig. 4A). Furthermore, after separation of the cells the IL12p35 subunit was dose dependently suppressed by rhIL6 on mRNA level in mDCs (up to 77.8%; Fig. 4B). Although IL12p19 was not affected by rhIL6, IL12p40 was reduced to a lesser extent (35% reduction) than p35 (Fig. 4C and D). IL6 neutralization in the conditioned media of SW756 and SiHa cells, which express up to 3586 pg/mL of IL6 (Fig. 4E), completely restored the IL12 expression of cervical cancer–instructed mDCs cocultured with NECFs (Fig. 4F), whereas the IL23 expression was not affected (Fig. 4G). Taken together, our results clearly showed that IL6 is the crucial factor for the cervical cancer cell–mediated IL12 suppression in cancer-instructed mDCs.

Next, we were interested in the mechanism of increased IL23 expression in mDCs mediated by cervical cancer–instructed fibroblasts. Stimulation of NECFs with conditioned media from different cervical cancer cells (a scheme of the stimulation experiments is presented in Fig. 5A) resulted in up to 8.7-fold enhanced expression of the fibroblast activation marker αSMA (Supplementary Fig. S1D) and lead to potent induction (up to 9.4-fold; Fig. 5B) of the cytokine IL1β, a potential inducer of IL23 expression (29), reaching levels up to 555 pg/mL (Fig. 5C). No significant induction was observed after stimulation with supernatants derived from normal HPV-negative NECK.

We have recently shown that cervical cancer cell–derived IL6 induces the C/EBPβ pathway in cervical fibroblasts (9), a regulator of IL1β expression (30, 31). NECFs expressed the IL6 receptor (IL6R) and rhIL6 induced IL1β production and C/EBPβ expression in NECFs (Supplementary Fig. S5A–S5C). To elucidate the impact of IL6/C/EBPβ-dependent signaling on IL1β induction, IL6-neutralizing antibodies and C/EBPβ-specific siRNAs were used. IL6 neutralization in the conditioned media of cervical cancer cells reduced C/EBPβ expression (Supplementary Fig. S5D) and led to a significant reduction of IL1β protein production by cervical cancer–instructed NECFs (77.8%–80.5% reduction; Fig. 5D). Furthermore, knockdown of C/EBPβ, which was confirmed by Western blot analysis (Fig. 5E; 77.9% reduction), significantly reduced IL1β production induced by cervical cancer cells in NECFs (Fig. 5F; up to 81.1% reduction). To analyze the relevance of IL1β produced by cervical cancer–instructed fibroblasts for the IL23 induction of mDCs, prestimulated fibroblasts were incubated with neutralizing IL1β antibodies. Neutralization of IL1β significantly reduced the IL23 expression in cocultures with normal and cervical cancer–instructed mDCs (Fig. 5G; 76% reduction).

CD83⁺ mDCs were in contact with stromal αSMA–positive myofibroblasts in cervical cancer tissues (Fig. 2B). To evaluate the impact of freshly isolated fibroblasts from cervical cancer outgrowth cultures ex vivo on the IL23 expression of mDCs, we generated DCs in the presence of fibroblast supernatants for 6 days (fibroblast-instructed DCs) or cervical cancer–instructed DCs. Freshly isolated cancer-instructed fibroblasts from three different donors released significantly higher amounts of IL6 (3,951 pg/mL; Fig. 5H) and IL1β (537 pg/mL; Fig. 5I) protein in their supernatants in comparison with NECFs (55 pg/mL or 19.5 pg/mL, respectively). Neither IL6 neutralization in the conditioned media used for generation of fibroblasts-instructed or cervical cancer–instructed DCs nor during cocultures with fibroblasts from cervical cancer outgrowth cultures did affect the IL23 production by mDCs (Fig. 5J). However, neutralization of IL1β during coculture of fibroblast-instructed DCs and fibroblasts from cervical cancer outgrowth cultures significantly reduced IL23 production by mDCs (Fig. 5K). In contrast, supernatants of fibroblasts from cervical cancer outgrowth cultures reduced the IL12 expression in mDCs and neutralization of IL6 in fibroblast supernatants significantly restored the IL12 expression (Supplementary Fig. S5E).

Taken together, our data clearly showed that cervical cancer cell- and cancer-associated fibroblast-derived IL6 is a crucial factor mediating the reduced IL12 expression in cervical cancer- and fibroblast-instructed DCs. Furthermore, cervical cancer cell–derived IL6 induced C/EBPβ-dependent IL1β expression in NECFs. IL1β produced by cancer-instructed NECFs or by primary fibroblasts from cervical cancer outgrowth cultures mediated the enhanced IL23 production in cocultured normal and cervical cancer–instructed mDCs.

To validate our in vitro results in vivo, we stained SCC biopsies with a pancytokeratin antibody and for IL1β and αSMA. Indeed, our in situ analysis revealed αSMA–positive myofibroblasts (green) in the stroma that coexpressed IL1β (red; Fig. 5L, top) and CD83-positive cells (green) coexpressing IL23p19 (red) in the same area of the tumor biopsy (bottom) confirming our in vitro results.

Next, we studied the impact of IL23 produced by cervical cancer–instructed mDC–fibroblast cocultures on the expansion of Th17 cells. CD4⁺ memory T cells were cultured with αCD2/CD3/CD28 beads in the presence of supernatants from mDC–fibroblast cocultures for 6 days (Fig. 6A). Intracellular flow cytometry analysis revealed increased frequencies of IL17-producing CD4⁺ memory T cells under the influence of rhIL23 or supernatants of normal (Fig. 6B, top) as well as cervical cancer–instructed mDC–NECF cocultures (Fig. 6B, bottom). CD4⁺/IFNγ⁺ (Th1) cells were only slightly affected and IL17⁺/IFNγ⁺ double-positive CD4⁺ T cells were not affected by coculture supernatants (Fig. 6C). Using Trucount Absolute Counting Tubes, we found enhanced Th17 numbers (up to 2.5-fold increase by co-mDC–NECF culture supernatants in comparison with medium control), which were further enhanced (up to 2.1-fold) by supernatants from normal or cervical cancer–instructed mDCs cocultured with prestimulated fibroblasts (Fig. 6D). Furthermore, we found significantly elevated IL17A protein levels (up to 3-fold) in the supernatants of CD4⁺ memory T cells cultured with rhIL23 or mDC–fibroblast supernatants in comparison with medium control (normalized to 1,000 Th17 cells; Fig. 6E). Highest IL17A amounts were produced by CD4⁺ memory T cells under the influence of coculture supernatants from mDCs and prestimulated fibroblasts (up to 2,108 pg/mL). Thus, our results show that cervical cancer–instructed mDC–fibroblast cocultures enhanced both the numbers of Th17 cells as well as the IL17A production per T cell.

Supernatants from mDC–fibroblast cocultures or rhIL23 slightly induced the IL17 expression in CD4⁺ naïve T cells (1.6-fold; Supplementary Fig. S6A). Supernatants of NECF and normal mDC cocultures or rhIL12 as a control induced IFNγ production in naïve CD4⁺ T cells after 6 days of culture (Supplementary Fig. S6B; 28% double-positive cells), but in naïve CD4⁺ T cells cultured in supernatants from NECF and cervical cancer–instructed mDC cocultures, the IFNγ production was significantly reduced (64.5% reduction).

To analyze the relevance of IL23 from mDC–fibroblast cocultures for Th17 expansion, we neutralized IL23 in supernatants of cocultures with normal as well as cervical cancer–instructed mDCs and fibroblasts. Anti-IL23 antibodies significantly reduced the frequencies (76% decrease; Fig. 6F) and numbers of Th17 cells (85.2% decrease; Fig. 6G) as wells as IL17A protein production by Th17 cells (96.7% reduction; Fig. 6H), which was not observed with the respective isotype control antibodies.

To investigate the relevance of IL1β produced by stimulated fibroblasts or ex vivo by fibroblasts from cervical cancer outgrowth cultures for the enhanced IL23 production in DCs and subsequent IL23-dependent Th17 expansion, we neutralized IL1β in fibroblast supernatants. Neutralization of fibroblast-derived IL1β and not CCL20 (Supplementary Fig. S6C–S6F) resulted in significantly reduced Th17 frequencies (Fig. 6I) and numbers (Fig. 6J) and per T-cell IL17A protein production (Fig. 6K). In conclusion, our data showed that cervical cancer–instructed mDC–fibroblast cocultures support IL17 production of CD4⁺ memory T cell in an IL23-dependent manner.

Th17 cells, a T-cell subset with protumorigenic properties, infiltrate cervical cancers and correlate with advanced tumor stages (9, 32). We previously described that cervical cancer cells actively contribute to Th17 cell recruitment during cervical cancer progression, instructing cervical fibroblasts to produce the Th17-chemoattractant protein CCL20 (9).

In this study, we investigated a novel mechanism underlying Th17 cell expansion. Our in situ analysis identified CD83⁺ mDCs in cervical cancer tissues expressing the IL23 subunit p19. Notably, CD83⁺IL23p19⁺ cells significantly correlated with stromal Th17 cells and the severity of the disease. We show that cocultivation of cervical cancer–instructed mDCs with cervical fibroblasts enhanced the IL23 expression of mDCs and prestimulation of fibroblasts with conditioned media of cervical cancer cells further increased the IL23 expression. In contrast, the IL12 expression of cervical cancer–instructed mDCs was strongly reduced. We identified cervical cancer cell-derived IL6 as the central mediator of the differential expression of IL23 and IL12. We show that cervical cancer cells induced C/EBPβ-dependent IL1β expression via IL6 in fibroblasts. We identified IL1β as the inducer of the IL23 expression of mDCs in mDC–fibroblast cocultures. Freshly isolated cancer-instructed fibroblasts supported the increased IL23 and reduced IL12 expression of mDCs by secretion of IL6 and IL1β ex vivo. In line with the previous results, IL6 suppressed IL12 production by targeting the IL12p35 subunit in mDCs (33). As a consequence, supernatants from cervical cancer–instructed mDC–fibroblast cocultures promoted the IL17 production in CD4⁺ memory T cells in an IL23-dependent manner but failed to induce Th1 differentiation. Figure 7 summarizes our current concept of our findings. This is the first report on a mechanism how the cervical cancer micromilieu can support Th17 expansion.

Th17 cells infiltrate the precursor lesions and their number further increase during cervical cancer progression (9, 14). So far, it was unclear how the Th17 expansion occurs in cervical cancers. IL23 has been shown to be necessary for Th17 expansion (20) and imbalances of DC-derived IL23 and IL12 expression in cancers favored Th17 polarization (34). IL23 exhibits protumorigenic properties directly by promoting tumor growth (35) or indirectly by supporting Th17 cell responses (20). Th17 cells have a paradox role in different cancers. They mediate antitumor effects by recruiting immune cells into tumors or stimulating effector CD8⁺ T cells (36, 37) as well as tumor-promoting responses driving proliferation, invasion, metastasis, and angiogenesis (38, 39). Patients with cervical cancer show increased serum levels of IL23, which correlated with their Th17 numbers in the blood (22). In this study, our in situ analysis showed CD83⁺ mDCs coexpressing IL23p19 in cervical cancer tissues. Notably, the number of CD83⁺IL23p19⁺ cells correlated with the numbers of Th17 cells. This correlation was stronger than the correlation between numbers of total IL23p19⁺ or CD83⁻ IL23p19⁺ cells and Th17 cells, underlining the relevance of DC-derived IL23. Furthermore, patients with advanced tumor stages and lymph node metastasis showed increased proportions of CD83⁺IL23p19⁺ cells per total CD83⁺ cells and CD4⁺IL17⁺ cells per total CD4⁺ cells. Thus, the number of IL23-expressing mDCs as well as IL17-producing CD4⁺ T cells apparently increases during later stages of carcinogenesis and is associated with the severity of the disease.

In our recent studies, we observed that cervical cancer cells induce the chemokine CCL20 in cervical fibroblasts supporting Th17 cell recruitment (9). Via the expression of proinflammatory cytokines and chemokines, such as IL1β and CCL20 (9, 40), fibroblasts can promote inflammation during progression of different cancer types (41). In peripheral tissues, interactions between fibroblasts and DCs can modulate DC functions with respect to their ability to mature, migrate, and trigger appropriate adaptive immune responses (42). Depending on their anatomic region of origin, fibroblasts can transfer tissue-specific information to DCs (43, 44). Thus, DCs cocultured with renal fibroblasts showed decreased expression of CD80, CD83, CD86, and IL12 but maintained normal levels of IL23 and IL27 (45). Skin-derived fibroblasts did not influence maturation marker expression of mDCs but support their IL23 expression and, thereby, Th17 responses (24) also after irradiation (46). In our study, cervical fibroblasts did not interrupt efficient phenotypic maturation of mDCs. Furthermore, cocultivation of cervical fibroblasts with mDCs enhanced the IL23 expression in normal as well as cervical cancer–instructed mDCs. We identified IL1β as the responsible factor produced by cervical cancer–instructed fibroblasts mediating the increase of IL23 expression in cocultured mDCs. Our data show that cervical cancer cells instruct fibroblasts via IL6/C/EBPβ to produce IL1β, thus engaging the same pathway previously described for the chemokine CCL20 (9). Although fibroblast-derived CCL20 supported the recruitment of Th17 cells (9), our neutralization experiments clearly demonstrated that CCL20 was not involved in the expansion. Rather, fibroblast-derived IL1β led to the IL23-dependent expansion and enhanced IL17 production of Th17 cells. Although the IL1β expression declines in HPV-infected epithelial cells during cervical carcinogenesis (47), we found stromal αSMA–positive activated myofibroblasts in situ coexpressing IL1β in cervical cancer biopsies. Notably, in consecutive sections of the biopsy, we also detected CD83⁺IL23p19⁺ cells in the same area of the tumor underlining our in vitro results.

IL23 is a heterodimeric cytokine mainly expressed by DCs containing the subunit IL23p19 and shares the subunit IL12p40 with IL12 (20). Apart from IL12p40, the cytokine IL12 consists of the subunit IL12p35, which is regulated by the transcription factor NF-κB (28). We had previously described that recombinant IL6 (48) or cervical cancer cell–derived IL6 suppresses the transcription factor NF-κB in mDCs (8). This resulted in reduced expression of the chemokine receptor CCR7 in cancer-instructed mDCs (8), reducing their susceptibility for lymph node homing signals. Here, we found that cervical cancer cell–derived IL6 dissociated the IL12 from IL23 expression in cancer-instructed mDCs by suppression of the IL12p35. Cervical cancer cell–derived IL6 suppressed IL12 production in mDCs in the absence or presence of fibroblasts. In contrast to p35, the IL23p19 subunit is mainly regulated by MAPK p38 (49). IL23p19 was not affected by IL6, which is in line with our previous studies demonstrating that recombinant IL6 did not affect p38 activation in mDCs (48). IL6 is highly upregulated during cervical carcinogenesis in situ and in vitro (12) and correlated with poor prognosis of the patients (10, 50). Cervical cancer cells display only low responses to autocrine IL6 due to low gp80 expression levels (12, 51). According to our previous studies (6, 8, 9, 11), our results support the notion that IL6 rather acts in a paracrine way to create a protumorigenic microenvironment. Thus, cervical cancer cell–derived IL6-instructed mDC–fibroblast cocultures mediated the expansion of Th17 cells in an IL23-dependent manner and reduced Th1 differentiation in vitro. By the secretion of IL6 and IL1β, cancer-instructed fibroblasts supported the underlying diverse IL23 and IL12 production in cocultured mDCs. Our study shows that IL23 is an important mediator of Th17 expansion that may act in concert with other factors of mDC–fibroblast supernatants. This differential occurrence of Th17 and Th1 cells and their related cytokines is described for patients with cervical cancer. Patients exhibit elevated numbers of Th17 cells and increased IL23 and IL17 but reduced IFNγ levels in their serum (22) and decreased numbers of HPV-specific Th1 cells as well (52).

Clinically most important, retrospective analysis from this study further supported the in vivo relevance of our findings. We found that patients with increased numbers of CD83⁺IL23p19⁺ cells and Th17 cells developed lymph node metastasis and recurrent cervical cancers. This may have major implications for personalization of cervical cancer therapy, suggesting the evaluation of the percentage of IL23p19-expressing CD83⁺ cells per total tumor-infiltrating CD83⁺ cells and proportion of IL17-expressing CD4⁺ cells per total tumor-infiltrating CD4⁺ cells as potential prognostic markers or targets for immunotherapy. Inhibitors of IL6 or of IL6 signaling are currently being evaluated for cancer treatment (53) that may target the IL6/C/EBPβ/CCL20-dependent recruitment of tumor-promoting Th17 cells to cervical cancer tissues (9). In mice, CCR6-mediated T-cell migration has been prevented using an engineered CCL20 variant (54). Inhibitors of IL6 may further target the expression of IL23 in DCs and subsequent Th17 expansion. In contrast, Th1 cells expressing IFNγ might be reconstituted, because IL12, a major regulator of Th1 differentiation, is suppressed by IL6. However, IL6 inhibitors may prevent IL6/STAT3-mediated sensitization of cervical cancer cells towards chemotherapeutic drugs (51) and should, therefore, not be applied prior to irradiation or chemoradiotherapy. Furthermore, antibodies against IL23p19, IL17A, or IL17 receptor are currently being evaluated for treatment of psoriasis, inflammatory diseases, and different cancers (55) and should also be considered for cervical cancer therapy.

In conclusion, our study identified a novel role of cervical cancer cell–derived IL6 and cervical fibroblasts with regard to Th17 cells in cervical cancers. Cervical fibroblasts not only support Th17 infiltration via IL6-mediated expression of the Th17-chemoatractant CCL20 (9), they also promote subsequent expansion of these protumorigenic cells via secreted IL6 and IL1β by enhancing the expression of IL23 in cervical cancer–instructed mDCs. The number of IL23-expressing DCs increases with severity of disease and may explain the accumulation of Th17 cells in the stroma of advanced cervical cancers, which may further support and maintain a chronic protumorigenic inflammatory micromilieu, cervical cancer progression, and may influence cervical cancer therapy.

E.-F. Solomayer has received speakers bureau honoraria from Pharma and has provided expert testimony, but not directly for this manuscript. No potential conflicts of interest were disclosed by the other authors.

Conception and design: B. Walch-Rückheim, S. Hegde, S. Smola

Development of methodology: B. Walch-Rückheim, L. Theobald, S. Hegde, S. Smola

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Walch-Rückheim, R. Ströder, L. Theobald, S. Hegde, R.M. Bohle, I. Juhasz-Böss, E.-F. Solomayer, S. Smola

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Walch-Rückheim, L. Theobald, S. Smola

Writing, review, and/or revision of the manuscript: B. Walch-Rückheim, R. Ströder, Y.-J. Kim, I. Juhasz-Böss, E.-F. Solomayer, S. Smola

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Walch-Rückheim, J. Pahne-Zeppenfeld, Y.-J. Kim, R.M. Bohle, I. Juhasz-Böss

Study supervision: B. Walch-Rückheim, E.-F. Solomayer, S. Smola

Other (conducting, analysis, and interpretation of histopathologic studies): Y.-J. Kim

The authors thank Tanja Tänzer and Birgit Glombitza for excellent technical assistance and all blood donors for providing their blood. This work was supported by a grant of the Saarland University (B. Walch-Rückheim). B. Walch-Rückheim received funding from the “Else Kröner-Fresenius-Stiftung” (2017-A64).

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.