Plasmacytoid Dendritic Cells Play a Key Role in Tumor Progression in Lipopolysaccharide-Stimulated Lung Tumor–Bearing Mice

Toll-like receptors play a key role in the innate immune system. The recognition of specific pathogen-derived (pathogen-associated molecular patterns) or endogenous (damage-associated molecular patterns) molecules by TLRs lead to the activation of the innate immune system and provide a bridge to the adaptive immune system (1). Although increasing evidence has highlighted the involvement of TLRs in cancer (2), these data remain controversial. Thus, the goal of our study was to more definitively understand the role of TLRs in lung cancer and dissect the cellular and molecular mechanisms involved. Lung carcinoma, one of the leading causes of death worldwide, is a poorly immunogenic cancer, resistant to the surveillance of the immune system. However, TLR4 is highly expressed on human lung cancer samples, but its role in the progression of lung cancer is not well understood.

In the 19th century, Dr. William Coley discovered that repeated injections of bacterial toxins (later identified as LPS/Coley’s toxin) could serve as an efficient antitumor adjuvant (3). Since then, numerous studies have been conducted to investigate the effects of LPS on tumors. Several studies on tumor cell lines showed that TLR4 stimulation facilitates cell proliferation (4, 5). Similarly, LPS facilitates carcinogenesis in mouse models of skin cancer (6), lung cancer (7), breast cancer (8), and colon cancer (9).

LPS elicits airway inflammation via the activation of TLR4 in a MyD88-dependent manner (10). However, the dose of LPS is of essential importance because low levels of LPS can induce a Th2-like immunity, whereas high levels of LPS induce Th1 responses in a mouse model of airway sensitization (11). The mechanism by which LPS signaling results in Th2 sensitization involves the activation of Ag-containing dendritic cells (DCs) (12). Myeloid DCs are well known to be involved in TLR4-mediated airway inflammation and the dose of LPS is responsible for Th1 or Th2 polarization (13). Little is known, however, on the role of plasmacytoid DCs (pDCs) in the lung when they are not directly stimulated with a specific TLR7 or TLR9 ligand. In our previous study we showed that the injection of CpG- oligodeoxynucleotide (ODN), TLR9 ligand, in tumor-bearing mice increased the influx of pDCs to the lung where they participated in tumor progression (14). In the present study we wanted to investigate the role of pDCs, even though they express no or very weak levels of TLR4. We observed that injection of LPS differently modulated tumor progression in a dose-dependent manner. A low dose of LPS increased lung tumor burden, whereas a high dose of LPS decreased tumor progression in the lungs. In both groups of mice, pDCs were recruited to the lung but they displayed different phenotype and activity. pDCs in low-dose LPS mice were tolerogenic whereas those recruited in response to high-dose LPS were essential for tumor regression, providing a potential mechanism for the dose-dependent effect of LPS that we observed.

Female specific pathogen-free C57BL/6J mice (6–8 wk of age; Harlan Laboratories, Udine, Italy) and athymic nude-Foxn1^nu mice (Harlan Laboratories) were fed a standard chow diet and housed under specific pathogen-free conditions at the Istituto Nazionale Tumori, Fondazione “G. Pascale.” All animal experiments were performed under protocols that followed the Italian and European Community Council for Animal Care (decree law no. 116/92).

Metastatic melanoma cells B16-F10 were purchased from the American Type Culture Collection and cultured in DMEM supplemented with 10% FBS, l-glutamine (2 mM), penicillin (100 U/ml), and streptomycin (100 μg/ml) (Sigma-Aldrich, Milan, Italy) in an atmosphere of 5% CO₂ at 37°C.

Mice were injected i.v. with 1 × 10⁵ B16-F10 cells (day 0) and 7 d later LPS (ultrapure LPS from Escherichia coli E0111:B4; Vinci Biochem, Florence, Italy) (0.1–10 μg/mouse) was administered by the i.p. route. LPS was injected once and mice were sacrificed at day 3 or 7. Lung, spleen, and mediastinal lymph nodes were isolated.

In some experiments, pDC-depleting Ab (m927 Ab [rat IgG], 500 μg/mouse; i.p.) (14) was administered on day 7 or on days 9, 11, and 13 before mice were sacrificed. The depleting Ab was injected i.p. on days 7 and 9 for experiments at 3 d after LPS treatment and on days 7, 9, 11, and 13 for experiments at 7 d after LPS treatment. pDCs were depleted the same day as LPS or PBS treatment because we observed a higher influx of pDCs into the lymph nodes and spleens 24 h after LPS administration (data not shown). m927 Ab depleted lung pDCs by ∼95% compared with rat IgG (14). m927 Ab was injected every 2 d, as the turnover of pDCs into the lung was ∼48 h (data not shown) (14). pDC-depleting m927 Ab was provided by Dr. Marco Colonna (Washington University, St. Louis, MO).

In another set of experiments, pDCs previously isolated from the lungs of tumor-bearing mice were adoptively transferred into recipient mice by i.v. injection (0.5 × 10⁶ cells/mouse in 100 μl PBS) once only. Seven days after the adoptive transfer the mice were sacrificed.

Lungs, lymph nodes, and spleens were isolated and digested with 1–0.5 U/ml collagenase (Sigma-Aldrich). Cell suspensions were passed through 70-μm cell strainers, and RBCs were lysed. Cell suspensions were used for flow cytometric analysis of different cell subtypes. Bronchoalveolar lavage fluid (BALF) was collected using 0.5 ml PBS containing 0.5 mM EDTA and cell counts were performed. Additionally, lungs were homogenized and cytokines measured.

pDCs (CD11c^lowB220⁺CD19⁻CD11b⁻ cells) were generated by digesting the lung from tumor-bearing mice after 3 d LPS (0.1–10 μg/mouse) or PBS treatment. Lungs were excised and digested with 1 U/ml collagenase (Sigma-Aldrich) and antibiotics for 45 min. Cell suspensions were passed through 70-μm cell strainers, and RBCs were lysed. For some experiments, pDCs were isolated using a negative selection for mouse pDCs (EasySep from StemCell Technologies/Voden Medical Instruments, Milan, Italy). Purity was checked by flow cytometry by using anti-CD11c, CD19, B220, and CD11b Abs (eBioscience, San Diego, CA) and was routinely ∼85–90% (Supplemental Fig. 4). In some experiments lung-derived pDCs were cultured (1 × 10⁵ cells/well) with T cells (ratio 1:10). CpG-ODN (type A; 0.01–10 μg/ml) was used to stimulate pDCs. Cell-free supernatant was tested for IL-6, TNF-α, IFN-α, IL-12p40, IL-10, and IL-1β release.

pDCs derived from the lungs of mice treated with LPS (0.1–10 μg/mouse) or PBS were isolated as previously described. Splenic CD8⁺ and CD4⁺ T cells were isolated using an EasySep stem cell kit (Voden Medical Instruments, Milan, Italy). These cells were isolated from C57BL/6J mice. Purity was checked by flow cytometry by using anti-CD3, CD19, CD4, and CD8 Abs (eBioscience) and was routinely ∼90% (15). pDCs and T cells were cultured for 24 h and 5 d. Cell-free supernatant was tested for IFN-γ and IL-10 release. CD4⁺ and CD8⁺ T cells were used for cell proliferation by using CFSE (eBioscience).

The composition of lung inflammatory cells was determined by flow cytometry (BD FACSCalibur; BD Biosciences, Milan, Italy) using the following Abs: CD11c-FITC, CD11b-PE-Cy5.5, Gr-1-PE, CD3-PE-Cy5.5, CD4-FITC or allophycocyanin, CD8-PE or allophycocyanin, F4/80-PE, B220-PE, CD19-PE-Cy5.5, Siglec-H-PE-Cy5.5, MHC class II-PE or allophycocyanin, MHC class I-FITC or allophycocyanin, CD80-PE, CD25-PE, Foxp3-PE-Cy5.5 (eBioscience).

TNF-α, IL-6, IL-12p40, IL-10, IFN-α, IFN-γ, IL-17A, and IL-1β were measured in homogenate and BALF by using commercially available ELISAs (eBioscience; R&D Systems, London, U.K.).

Left lung lobes were fixed in OCT medium (Pella, Milan, Italy) and 7-μm cryosections were cut. H&E staining was performed and used to measure the tumor burden. Tumor lesions were analyzed by using serial lung cryosections and expressed as lung tumor foci count, as brown spots were counted from at least five serial sections (15).

Results are expressed as means ± SEM. Changes observed in treated groups compared with controls were analyzed using one-way ANOVA, followed by a Bonferroni posttest and/or Student t test. A p value <0.05 was considered significant.

To investigate the role of LPS in lung carcinoma we used a mouse model by which the metastatic B16-F10 cells were i.v. injected into the tail vein of C57BL/6J mice (Fig. 1A). B16-F10 cells preferentially localize to the lung where they appear as pigmented foci on the lung surface (Fig. 1C) (15). Seven days after B16-F10 cell inoculation, low- or high-dose LPS (0.1 or 10 μg/mouse) was administered i.p. Mice were sacrificed 3 or 7 d after the single administration of LPS at days 10 and 14, respectively (Fig. 1A).

The administration of low-dose LPS (0.1 μg/mouse) significantly increased pulmonary metastasis in tumor-bearing mice compared with PBS-treated mice (PBS, 86.57 ± 10.77 versus LPS [0.1 μg/mouse], 134.3 ± 17.15) (Fig. 1B, 1C). In contrast, injection of high-dose LPS (10 μg/mouse) not only did not increase the amount of lung tumor foci but instead tended to decrease such foci (although not in a significant manner) compared with PBS (Fig. 1B, 1C) (PBS, 86.57 ± 10.77 versus LPS [10 μg/mouse], 68.56 ± 13.07). This effect was observed after 7 d LPS or PBS treatment (Fig. 1B, 1C) but it was also visible after 3 d LPS or PBS treatment (PBS, 77 ± 9.83; LPS [0.1 μg/mouse], 124.5 ± 17.9; LPS [10 μg/mouse], 87.67 ± 9.17). To determine whether LPS induced lung inflammation in tumor-bearing mice as in normal mice (16), we evaluated the number of cells in the BALF of PBS- and LPS-treated mice. BALF cell numbers were lower in PBS- and low-dose LPS–treated tumor-bearing mice (PBS, 1.89 ± 0.84 × 10⁶ cells/ml; LPS [0.1 μg/mouse], 1.57 ± 0.54 × 10⁶ cells/ml) compared with high-dose LPS–treated tumor-bearing mice (9.15 ± 2.72 × 10⁶ cells/ml) (Supplemental Fig. 1A). These results indicate that tumor-bearing mice treated with high-dose LPS had an increased lung inflammation compared with low-dose LPS and PBS.

To begin to dissect the molecular mechanisms of our observations and to better understand the lung microenvironment, we first determined the levels of pro- and anti-inflammatory cytokines in the BALF. High-dose LPS increased the release of proinflammatory and cytotoxic cytokines IFN-γ (Fig. 1D) and IL-17A (Fig. 1E) in the BALF. In contrast, IL-10, an anti-inflammatory cytokine, was not affected in mice either treated with high-dose LPS or PBS. Conversely, low-dose LPS injection did not affect the levels of IFN-γ (Fig. 1D), but it did increase the levels of IL-17A (Fig. 1E) and, more importantly, it increased the release of IL-10 (Fig. 1F) in the BALF.

We next determined the identity of the cells recruited to the lung of tumor-bearing mice after LPS treatment. We digested the lungs and performed flow cytometry analysis to measure the influx of immune cell infiltrates. pDCs were determined at both 3 and 7 d after LPS treatment and were identified as CD11c^low, B220⁺, CD11b⁻, and Gr-1^int cells as previously reported (17). Surprisingly, the percentage of pDCs into the lung of LPS-treated tumor-bearing mice was increased in a dose-dependent manner compared with PBS-treated mice at day 10 (Fig. 1G). The same was observed at day 14 (Supplemental Fig. 1B). To confirm these results, lung pDCs were isolated. The number of pDCs in the lung of tumor-bearing mice was increased in a dose-dependent manner after LPS treatment (Fig. 1H).

The influx of conventional DCs (cDCs), identified as CD11c⁺CD11b⁺F4/80⁻, was higher in the lung of tumor-bearing mice treated with high-dose LPS compared with PBS and low-dose LPS (Supplemental Fig. 2A). High-dose LPS also induced activation of the cDCs as determined by expression of CD80 (Supplemental Fig. 2B), MHC class II (Supplemental Fig. 2C), and MHC class I (Supplemental Fig. 2D). However, low-dose LPS did not induce the activation of the cDCs to the same extent, as these cells had a higher MHC class I, but expressions of CD80 and MHC class II were not significantly changed from PBS controls (Supplemental Fig. 2B–D). These results indicate that the increased presence of pDCs in the lungs of mice treated with high-dose LPS was associated with activated cDCs, whereas pDCs recruited to the lung of mice treated with low-dose LPS was associated with less active or immature cDCs.

Our previous data showed that the higher recruitment of pDCs to the lung was associated with the increase of immunosuppressive cells, favoring tumor growth in the lung (14, 15). To evaluate whether LPS-mediated increase of pDCs to the lung of tumor-bearing mice could skew toward a suppressive immune environment, we analyzed the recruitment of myeloid-derived suppressor cells (MDSCs), regulatory T cells (Tregs), and CD8⁺ Tregs. MDSCs were identified as CD11b⁺Gr-1⁺, as previously published (15). Low-dose LPS resulted in greater influx of MDSCs to the lungs of tumor-bearing mice compared with PBS (Fig. 2A). High-dose LPS still resulted in a slight increase, although not significant, in MDSCs to the lung of tumor-bearing mice, but it was significantly reduced compared with low-dose LPS (Fig. 2A). Moreover, the influx of CD4⁺CD25⁺Foxp3⁺ cells (Tregs) was increased in the lungs of mice treated with low-dose LPS (Fig. 2B) compared with high-dose LPS and PBS controls. Furthermore, the percentage of CD3⁺CD8⁺ T cells was increased in a dose-dependent manner in the lungs of tumor-bearing mice after LPS treatment (Fig. 2C). However, low-dose LPS–induced CD8⁺ T cells presented a significant increase in Foxp3 staining compared with high-dose LPS and PBS controls (Fig. 2D), indicating a potential suppressive phenotype (15, 18, 19).

Taken together, these data indicate that low-dose LPS increases the influx of immunosuppressive cells to the lung of tumor-bearing mice compared with high-dose LPS administration. Intriguingly, pDCs were recruited to the lungs of tumor-bearing mice after both low- and high-dose LPS administration, but following low-dose LPS the pDCs recruited displayed a suppressive environment whereas pDCs recruited following high-dose LPS displayed a cytotoxic phenotype.

We next investigated whether these differential pDC phenotypes induced with low- versus high-dose LPS were associated with the differential LPS effects we observed on tumor outgrowth. pDCs can drive CD4⁺ T cells to become CD4⁺CD25⁺Foxp3⁺ cells, which are identified as Tregs (20, 21). The role of Tregs in a tumor microenvironment can be translated into anergy and immunosuppression, favoring the immune escape of tumor cells with the concomitant increase of tumor burden (14, 22). To evaluate whether pDCs were crucial for tumor outgrowth after LPS treatment, we depleted pDCs using a pDC depleting Ab (m927 Ab) (14, 23). m927 Ab or IgG control was injected the same day as LPS administration and every 2 d before mice were sacrificed on day 14 (7 d after LPS injection) (Fig. 3A). Supplemental Fig. 2E and 2F show that the depletion of pDCs was maintained even after the administration of LPS at days 9 (Supplemental Fig. 2E) and 14 (Supplemental Fig. 2F). The administration of IgG control did not alter the morphology of the lung previously observed when LPS was administered alone in B16-F10 implanted mice. In contrast, the injection of m927 Ab increased the influx of inflammatory cells to the lung of tumor-bearing mice after low-dose LPS (Supplemental Fig. 3), whereas it significantly reduced the amount of cells counted in the BALF harvested from mice treated with high-dose LPS (Supplemental Fig. 3). Of interest, the depletion of pDCs from mice treated with low-dose LPS significantly reduced lung tumor foci (Fig. 3A). In contrast, the depletion of pDCs from the lungs of mice treated with high-dose LPS favored tumor growth (Fig. 3A). Although the m927 Ab is fairly specific for pDCs, we wanted to confirm our observations that indeed pDCs were involved in this phenotype. Thus, we performed additional experiments where pDCs derived from the lungs of tumor-implanted mice were adoptively transferred into PBS-, LPS (0.1 μg/mouse)-, and LPS (10 μg/mouse)-treated tumor-bearing mice. The administration of low-dose LPS still increased lung tumor foci count in pDC adoptively transferred mice (Fig. 3B, dotted bars). The same was observed with the high-dose LPS that still induced tumor arrest in the lung of mice adoptively transferred with tumor-derived pDCs (Fig. 3B, dotted bars). Thus, these additional data support and confirm what we observed with the pDC-depleting Ab and better highlight that the immune microenvironment influences pDC phenotype. Furthermore, pDC depletion led to a decrease in MDSCs (Fig. 3C), Tregs (Fig. 3D), and CD8⁺ Tregs (Fig. 3F), whereas the percentage of Foxp3⁻CD8⁺ T cells increased in mice treated with low-dose LPS (Fig. 3E). Finally, pDC depletion in high-dose LPS–treated mice increased the influx of MDSCs (Fig. 3C) to the lungs and decreased the percentage of CD8⁺ T cells (Fig. 3E). In this latter case, we did not detect any difference between Tregs (Fig. 3D) and CD8⁺ Tregs (Fig. 3F) probably because we had a higher Th2-like immune environment as assessed by the release of IL-13 (Supplemental Fig. 1D).

To evaluate the tumor environment after the depletion of pDCs, we measured the expression of proinflammatory (IFN-γ and IL-17A) and anti-inflammatory (IL-10) cytokines in the BALF (Fig. 3). The ablation of pDCs in tumor-bearing mice during low-dose LPS treatment increased the release of IL-17A (Fig. 3G) and IFN-γ (Fig. 3H) and significantly reduced IL-10 (Fig. 3I). In contrast, IL-17A (Fig. 3G) and IFN-γ (Fig. 3H) were significantly reduced after high-dose LPS treatment in pDC-depleted tumor-bearing mice. However, the levels of IL-10 did not change (Fig. 3I) in high-dose LPS mice, and these data correlate with there being no increase in either CD4 or CD8 Tregs in these mice (Fig. 3D, 3F). Taken together, the data suggest that pDC depletion reverses the protumor environment that is initiated by low-dose LPS administration whereas during high-dose LPS, pDC depletion reverses the antitumor effects. These results imply that although TLR4, the LPS receptor, is not expressed (or very weakly expressed) on pDCs, the activation of the immune system in the lungs of tumor-bearing mice facilitates the recruitment of pDCs, which are crucial for both lung tumor progression and inhibition.

To understand the phenotype of pDCs into the lung of tumor-bearing mice treated with differing amounts of LPS, we isolated pDCs from the lungs of mice treated with PBS or LPS (0.1–10 μg/mouse) at day 10 by means of magnetic isolation, with a routine purity of 85–90% (Supplemental Fig. 4). pDCs isolated from tumor-bearing mice that were treated with low-dose LPS (0.1 μg/mouse) had no difference in Siglec-H expression compared with mice treated with PBS (Fig. 4A) but they presented a higher expression of ICOS ligand (ICOSL) (Fig. 4B) and programmed death ligand-1 (PD-L1) (Fig. 4C). In contrast, pDCs isolated from mice treated with high-dose LPS (10 μg/mouse) had lower levels of Siglec-H (Fig. 4A) and presented similar levels as did PBS-treated tumor-bearing mice for ICOSL (Fig. 4B) and PD-L1 (Fig. 4C). These data confirm the tolerogenic nature of pDCs after treatment of tumor-bearing mice with low-dose LPS compared with the high-dose LPS–induced pDCs.

To evaluate the Ag presentation capacity of pDCs, we analyzed the levels of MHC class I, MHC class II, and CD80. High-dose LPS (10 μg/mouse) significantly reduced the levels of MHC class I (Fig. 4D) and MHC class II (Fig. 4E) on pDCs compared with mice treated with PBS and low-dose LPS (0.1 μg/mouse). No statistical differences were observed for CD80 levels (Fig. 4F). Additionally, in vitro restimulation of low-dose LPS–derived pDCs with CpG-ODN, TLR9 ligand, favored TNF-α (Fig. 4G), IL-10 (Fig. 4H), and IL-6 (Fig. 4K) release. In contrast, high-dose LPS–derived pDCs produced higher levels of IL-12p40 when restimulated with CpG (Fig. 4L). No statistical differences were observed for IFN-α release (Fig. 4J) and IL-1β (Fig. 4I), although in this latter case we observed a higher baseline level (PBS) between pDCs derived from mice treated with PBS and low-dose LPS compared with pDCs derived from mice treated with the higher dose of LPS (Fig. 4I).

Our data show that the pDCs have differential phenotypes in the lungs of mice treated with either low- or high-dose LPS. pDCs appear tolerogenic during low-dose LPS administration whereas they are Th1-polarizing during high-dose LPS. To understand their role in T cell proliferation and polarization, we performed MLR experiments. The coculture of pDCs from mice treated with PBS, low-, or high-dose LPS with naive (obtained from the spleen of non–tumor-bearing mice) CD4⁺ T cells did not alter their proliferation rate (Fig. 5A), although CD4⁺ T cells became more activated (CD69⁺ cells) when they were cocultured with pDCs derived from mice treated with high-dose LPS (Fig. 5B). Additionally, pDCs derived from mice treated with low-dose LPS increased CD8⁺ T cell proliferation (Fig. 5C) but did not alter the expression of CD69 (Fig. 5D). In contrast, similar to CD4⁺ T cells, the coculture of pDCs derived from mice treated with high-dose LPS did not increase the proliferation rate of CD8⁺ T cells (Fig. 5C) but increased CD69 levels on CD8⁺ T cells (Fig. 5D). To understand how pDCs affected T cell polarization, cell-free supernatant was tested for Th1-like (IFN-γ) and immunosuppressive (IL-10) cytokines. The levels of IL-10 greatly increased from the coculture of CD4⁺ T cells and pDCs derived from mice treated with low-dose LPS (Fig. 5E) compared with the coculture of pDCs from PBS- and high-dose LPS–treated tumor-bearing mice (Fig. 5E). Similar results were obtained using CD8⁺ T cells (Fig. 5F). Conversely, IFN-γ was increased in the supernatant from the coculture of naive CD4⁺ (Fig. 5G) and CD8⁺ T cells (Fig. 5H) with pDCs derived from high-dose LPS–treated tumor-bearing mice and not low-dose LPS. Moreover, the coculture of naive lung-derived pDCs with naive CD4⁺ or CD8⁺ T cells did not alter the proliferation rate (Fig. 5A, 5B), the activation (Fig. 5C, 5D), and the release of IL-10 (Fig. 5E, 5F) and IFN-γ (Fig. 5G, 5H) from CD4⁺ or CD8⁺ T cells.

Taken together, our data indicate that the differential activity of pDCs in mice treated with the low dose and high dose of LPS is due to their phenotype, which is immunosuppressive in the case of tumor-bearing mice that were treated with low-dose LPS and Th1-polarizing when tumor-bearing mice were treated with high-dose LPS.

Several investigators reported that pDCs either do not express TLR4 (24, 25) or they express very weak TLR4 (26–28). Thus, to understand how LPS, a TLR4 ligand, could induce the recruitment of pDCs to the lung of tumor-bearing mice, we used athymic nude mice, which lack T cells. Nude mice were inoculated with B16-F10 cells in the same manner as described for immunocompetent C57BL/6J mice (Fig. 1A). Tumor-bearing nude mice treated with LPS presented a higher count of lung tumor foci when treated with high-dose LPS compared with PBS and a low dose of LPS (Fig. 6A). Surprisingly, the recruitment of pDCs to the lungs of nude mice was reduced with both the low- and high-dose LPS (Fig. 6B) compared with PBS and C57BL/6J controls. To understand what T cell component was responsible for pDC recruitment, we reconstituted tumor-bearing nude mice with CD4⁺ or CD8⁺ T cells at the same time LPS was injected. Naive CD4⁺ or CD8⁺ T cells were i.v. inoculated. The inoculation of CD4⁺ T cells did not alter lung tumor foci count after LPS treatment in nude mice (Fig. 6C, dotted bars) compared with C57BL/6J mice (Fig. 6C, open bars). However, tumor-bearing nude mice treated with high-dose LPS had a higher percentage of pDCs to the lung compared with nude mice treated with PBS or low dose of LPS and to C57BL/6J mice (Fig. 6D, open bars). Similarly, the inoculation of CD8⁺ T cells to tumor-bearing nude mice (Fig. 6E, dotted bars) did not alter the foci count of the pulmonary metastases compared with C57BL/6J mice (Fig. 6E, open bars), although the treatment with low-dose LPS led to higher amounts of lung metastases compared with mice that were inoculated with CD4⁺ T cells (Fig. 6C) and with nude mice (Fig. 6A). The inoculation of CD8⁺ T cells significantly increased the percentage of pDCs to the lung of nude mice (dotted bars) treated with low-dose LPS compared with PBS and high-dose LPS (Fig. 6F, open bars) and to C57BL/6J mice. These data imply that CD4⁺ T cells are responsible for pDC recruitment to the lung of tumor-bearing mice treated with high-dose LPS, whereas CD8⁺ T cells are responsible for the recruitment of pDCs to the lung of tumor-bearing mice treated with low-dose LPS.

The role of pDCs in modulating tumor growth is an actively investigated area of research. pDCs have been found in many human solid tumors, including lung cancer, and correlates to a poor patient prognosis (29). They are present in their nonactivated state and have been associated with the development and maintenance of the immunosuppressive tumor microenvironment (30). In this study we found that the higher tumor burden in the lung of low-dose LPS–treated tumor-bearing mice was dependent on the presence of pDCs, which had a tolerogenic phenotype. The ablation of pDCs decreased the immunosuppressive environment in the lung of low-dose LPS–treated tumor-bearing mice, facilitating tumor regression. In contrast, pDCs derived from the lung of tumor-bearing mice treated with high-dose LPS facilitated Th1 and cytotoxic T cell bias with the concomitant arrest of tumor growth. Although TLR4 is not or weakly expressed by pDCs, we think that the initial activation of the innate immune system via TLR4 facilitated pDC recruitment to the lung. Indeed, the production of CXCL10 (IP-10), which underlies TLR4 signaling, was increased in the BALF of mice treated with LPS (Supplemental Fig. 1C). High-dose LPS increased cDC maturation/activation and favored a Th1- and Th17-like bias. In this scenario, pDCs were in their active phenotype and facilitated tumor regression. This was associated with increased percentage of NK (Supplemental Fig. 1E), NKT (Supplemental Fig. 1F), and cytotoxic T cells. Indeed, the ablation of pDCs reversed this condition and facilitated tumor outgrowth. In contrast, low-dose LPS favored a Th2-like environment as demonstrated by higher levels of IL-13 (Supplemental Fig. 1D) and an immunosuppressive environment, as demonstrated by IL-10 levels and suppressive immune cells. The higher expression of ICOSL and PD-L1 on pDCs derived from mice treated with low-dose LPS is a hallmark for their suppressive/inactive phenotype. In support, high levels of ICOSL promote survival, expansion, and IL-10–producing Tregs (31), converting Ag-specific T cell proliferation (32). Similarly, PD-L1 facilitates cell contact between pDCs and Tregs (22). Thus, our data support the finding that nonactivated pDCs (with an immature phenotype) can favor latent chronic inflammation induced by low-dose LPS in the lung of tumor-bearing mice with the concomitant progression of tumor cell proliferation. Low-dose LPS–treated tumor-bearing mice presented a higher immunosuppressive environment characterized by higher recruitment of Tregs, MDSCs, and CD8⁺ Tregs compared with the high-dose LPS group. However, the exact roles of pDCs on CD8⁺ T cells, which were unable to mount a CTL reaction owing to their higher expression of Foxp3, still remain to be elucidated. The coculture experiments of naive CD8⁺ T cells with pDCs derived from the lung of tumor-bearing mice treated with low-dose LPS showed higher release of IL-10, implying the suppressive nature of CD8⁺ T cells (15). However, pDCs obtained from mice treated with low-dose LPS also produced higher levels of IL-10 but only after CpG treatment (Fig. 4H). Thus, we could speculate that the tolerogenic pDCs rendered the naive CD8⁺ T cells into suppressive IL-10–producing cells. In support of this hypothesis, Gilliet and Liu (33) found that human pDCs could skew CD8⁺ T cells into IL-10–producing suppressive cells. Additionally, we observed that low-dose LPS–derived pDCs increased the cell proliferation rate of CD8⁺ T cells, although they were not active.

Loschko et al. (34) showed that the recognition of Ags by Siglec-H dampens pDC-induced adaptive responses. In our study, Siglec-H was expressed on pDCs derived from tumor-bearing mice treated with PBS or a low dose of LPS. Instead, high-dose LPS pDCs had reduced amounts of Siglec-H. Although Siglec-H was described as a specific marker for pDCs, expression of it can discriminate among proinflammatory or tolerogenic pDCs (34). In support of this, the higher expression of Siglec-H on lung-derived pDCs obtained from low-dose LPS but not from high-dose LPS group correlated with the higher presence of immunosuppressive cells and cytokines that facilitate tumor development

pDCs have a limited Ag presentation capability (30). In our experimental conditions we did not observe an increase in MHC class I, MHC class II, and CD80 on pDCs derived from the lung of tumor-bearing mice treated with LPS. Instead, we detected lower expression of these markers on pDCs isolated from the lung of tumor-bearing mice treated with high-dose LPS. Most probably, this was due to the absence of Ag recognized by pDCs. LPS is “sensed” by TLR4, which is weakly or not expressed by pDCs. Thus, we postulate that the involvement of pDCs is secondary to the innate immune arm activated by TLR4 signaling. In the same context, however, pDCs were essential to inhibit tumor progression in mice treated with high-dose LPS. This could be explained by the presence of Th1-polarizing pDCs in these mice compared with PBS and low-dose LPS. We did not detect any difference in IFN type I basal levels from pDCs derived from mice treated with LPS compared with PBS. Although it was demonstrated that IFN type I can negatively regulate pDC turnover during inflammatory conditions (35), we did not observe similar results in our experimental conditions. Note that the stimulation of lung tumor–derived pDCs from PBS and high-dose LPS–treated tumor-bearing mice showed differential production of IFN type I after TLR9 stimulation with CpG. Thus, IFN type I production under stimulation was directly correlated to the phenotype of pDCs in the tumor. Alternatively, Drobits et al. (36) showed that TLR7-mediated activation of pDCs rendered these cells killer DCs able to directly eliminate tumor cells. This effect was similar to what we observed with high-dose LPS in lung tumor–bearing mice. In support of this concept, the inoculation of tumor-associated pDCs into tumor-bearing mice still induced high tumor foci count in low-dose but not high-dose LPS groups. In this study, we suggest that the phenotype of pDCs depends on the tumor microenvironment as it happens during infectious disorders (37). However, much work remains to be done to elucidate how the tumor microenvironment affects the pDC phenotype.

The recruitment of pDCs after LPS treatment was concomitant to the release of CXCL10 (IP-10), which binds to CXCR3, a receptor that is present on pDCs (30). In our conditions IP-10 was released in the BALF of mice treated with LPS. The synthesis/release of this cytokine is strongly induced in stromal cells upon type I IFN activity, suggesting both that pDCs self-sustained their recruitment to the inflamed lung of tumor-bearing mice and, most likely, that pDCs were recruited after the recognition of LPS by TLR4. The activation of TLR4 by LPS induces type I IFN release via a TRIF-dependent pathway (2, 3). IP-10 production is strictly dependent on IFN type I release in stromal and hematopoietic cells (2). In this context, our experimental conditions showed that the T cell component was essential for pDC recruitment to the lung of tumor-bearing mice. CD4⁺ T cells were required for pDCs after high-dose LPS, whereas CD8⁺ T cells were essential for pDC recruitment after low-dose LPS injection. When nude mice were reconstituted with CD4⁺ T cells, the pDC percentage was increased, although the amount of pulmonary metastases was still high, implying that CD8⁺ T cells collaborate in this scenario to diminish tumor growth after the injection of high-dose LPS. In contrast, CD8⁺ T cells were responsible for pDC recruitment after low-dose LPS, but again the amount of pulmonary metastases was still high, as demonstrated for C57BL/6J mice. In this latter case we observed that CD8⁺ T cells were not cytotoxic but were rendered immunosuppressive owing to the higher expression of Foxp3. Future studies will investigate whether the recruited pDCs or the local microenvironment rendered these cells suppressive. We hypothesize that the activation of the innate response by LPS does not lead to suppressive immune environment by itself; instead, it is the tumor microenvironment that disables an effective adoptive antitumor immune response. Our previous study showed that the treatment of mice with CpG-ODN, TLR9 ligand, increased the recruitment of pDCs associated with immunosuppressive tumor microenvironment due to the presence of Tregs and MDSCs (14, 15). CpG-ODN facilitated stromal cells to produce VEGF and FGF-2, which orchestrate tumor angiogenesis and suppressive immune environment (38). Similarly, tolerogenic pDCs facilitated tumor outgrowth in mice that were treated with low-dose LPS, but not in the case of high-dose LPS.

Collectively, our data show that the stimulation of TLR4 differentially modulates tumor outgrowth depending on the immune microenvironment generated by a low or high dose of LPS. Our study further sheds light on how a low dose of lung-sensitizing Ags such as LPS can facilitate tumor progression. Low levels of environmental LPS in inhaled air has been established to lead to latent chronic inflammation, which is deleterious for many pathologies that turn into a Th2-like immune environment (39–41). For example, asthma, which is characterized by a Th2-like environment (24), is exacerbated by low-dose LPS compared with high-dose LPS. Similarly, pDCs dampen lung inflammation into the lung of OVA-sensitized mice and in mice treated with cigarette extracts (10, 42), proving their suppressive immune activity. In our experimental conditions we observed a similar path, as the ablation of pDCs in tumor-bearing mice that were treated with low-dose LPS favored tumor arrest in contrast to mice treated with the high-dose LPS, implying a central role for pDCs in tumor progression in this model.

In conclusion, our data demonstrate that the dose of Ag and the tumor microenvironment determine whether pDCs alter the adaptive immunity in favor of tumor progression or regression. Humans are often exposed to low doses of air-circulating endotoxin, which can favor tumor microenvironment and cancerous cell proliferation. Alternatively, severe Gram-negative bacterial lung infections may provide the high local LPS condition and may be associated with tumor regression. The ablation of pDCs in our model of pulmonary metastasis also opens new therapeutic antitumor perspectives. Strategies seeking to modulate pDC numbers and activity in the tumor site might prove to be novel and effective as antitumor strategies.

The authors have no financial conflicts of interest.