IL-1β induces IDO1 expression in lung adenocarcinoma cells but not in normal lung epithelial cells

Since TDO2 was previously found to be upregulated concomitantly with IL-1β in normal lung cells [25], we asked whether IL-1β had the ability to regulate Trp-metabolizing enzymes in lung cancer cells. We chose 7 lung cancer cell lines based on the variable levels of expression of IDO1 and TDO2 (Fig. 1A). These cell lines harbor several different oncogenes representing a broad mutation profile that was observed in lung adenocarcinoma patients (described in Table 1). RNAseq data for the 7 lung cancer cell lines (Minna lab, dbGaP Study Accession phs001823.v1.p1), we determined that TDO2 mRNA was elevated in A549, H1792, H2347, and H2228 cells, while IDO1 and IDO2 were low in most cell lines (Fig. 1A). The normal immortalized lung epithelial cells HBEC-3KT did not express TDO2, IDO1, or IDO2 (Fig. 1A). We incubated HBEC-3KT and HSAEC-1KT, as well as the 7 lung adenocarcinoma cells of varying driver mutations (Table 1) with 5 ng/mL IL-1β for 48 h and performed RT-qPCR, Western blotting, and cell viability assays to determine their response to IL-1β (Fig. 1B). We found that IL-1β did not induce proliferation of the A549, H1792, HCC364 and HCC827 cell lines (Fig. 1C). In agreement with previous studies in breast and prostate cancers [34, 35, 37, 38], these results indicate that IL-1β does not promote cell-autonomous growth advantages to lung adenocarcinoma cells in vitro.

Fig. 1

Induction of IDO1 mRNA and protein levels in lung adenocarcinoma cell lines stimulated with IL-1β. A Heatmap showing the FPKM values of TDO2, IDO1, IDO2, CD274 and RelA mRNA of the cell lines used in this study. B Schematic of experimental set-up illustrating the cell lines tested. C Lung adenocarcinoma cells were treated with 5 ng/mL IL-1β, and cell viability measured 48 h after stimulation (n = 3 biological replicates). D HBEC-3KT and HSAEC-1KT “normal” lung epithelial cells or lung adenocarcinoma cell lines were grown in medium with vehicle control or 5 ng/mL IL-1β for 48 h, and RT-qPCR of IDO1 was performed. Note split scale for IDO1. E RT-qPCR showing induction of CXCL8 mRNA as a measure of activation of IL-1 signaling in HBEC-3KT and HSAEC-1KT cells in response to 5 ng/ml IL-1β treatment for 48 h. F HBEC-3KT and HSAEC-1KT cells show an induction of IDO1 mRNA in response to 5 ng/mL Ifnγ treatment for 48 h. G TDO2 mRNA levels measured by RT-qPCR in normal and lung adenocarcinoma cells in response to IL-1β treatment for 48 h. H IDO1 mRNA levels were measured by RT-qPCR in HCC827 cells treated with 0.1, 1, or 10 ng/mL of IL-1β for 48 h. I RT-qPCR of IDO1 mRNA in H2228 cells treated with 5 ng/mL IL-1β or 100 ng/mL IL-1Ra for 48 h. J Western blot analysis was performed for IDO1 and TDO2 protein levels in H1792 and HCC827 lung adenocarcinoma cells treated with vehicle control or 5 ng/mL IL-1β for 48 h. GAPDH was used as the loading control. K Cell counts were measured for monolayer cultures of H1792 and HCC827 cells stimulated with 5 ng/mL IL-1β ± 20 nM epacadostat (IDOi) for 4 days. Error bars represent ± SD of 3 biological replicates; *P ≤ 0.05, ** ≤ 0.005, *** ≤ 0.0005. mRNA levels were normalized to 18s mRNA and represented as fold change over HBEC-3KT control treatment. GAPDH was used as the loading control

Table 1 Lung adenocarcinoma cells and driver mutation

We then asked whether IL-1β activates pathways that lead to immune tolerance of lung cancer cells. We observed a significant upregulation of IDO1 mRNA (Fig. 1D) in A549, H1792, H1838, H2347, HCC364 and HCC827 adenocarcinoma cell lines incubated with IL-1β. Both HBEC-3KT and HSAEC-1KT cell lines displayed low expression of IDO1, and no additional induction was observed with IL-1β stimulation (Fig. 1D). Despite the lack of IDO1 induction, upon incubation with IL-1β, HBEC-3KT and HSAEC-1KT cells induced the expression of CXCL8 mRNA, a downstream target of IL-1 signaling [34, 39], indicating that these normal cells have the IL-1 receptors as well as an intact downstream signaling cascade in response to IL-1β stimulation (Fig. 1E). In addition, the mRNA expression of IL-1 receptors, IL-1R1 and IL-1RAcp (Minna lab dataset, dbGaP Study Accession is phs001823.v1.p1), and response to IL-1β stimulation, which was measured as fold change in CXCL8 induction, were detected in immortalized normal cells and lung adenocarcinoma cells (Table 1 and Supplemental Fig. 1A). To determine whether the IDO1 gene and promoter are intact in normal cells, we treated the cells with a known potent inducer, interferon gamma (Ifnγ), and found that Ifnγ could induce IDO1 mRNA (Fig. 1F). This indicates that normal cells can respond to cytokine stimuli to induce IDO1, but the IL-1β-mediated induction of IDO1 was specific to cancer cells.

Since both IDO1 and TDO2 can catalyze the conversion of Trp to Kyn, we asked whether IL-1β can also induce TDO2 expression in adenocarcinoma cells. Both HBEC-3KT and HSAEC-1KT cell lines had low TDO2 expression and showed no induction with IL-1β (Fig. 1G). Although A549 and H1792 cell lines had high expression of TDO2, none of the lung adenocarcinoma cells showed a significant induction with IL-1β stimulation (Fig. 1G). These results suggest that IL-1β specifically regulates IDO1 expression in lung adenocarcinoma cells.

IDO1 expression was IL-1β dose-dependent in the HCC827 cell line, with maximal expression observed at 1 ng/mL IL-1β (Fig. 1H). Interestingly, no further IDO1 mRNA induction with IL-1β occurred in H2228 cells, which had the highest IDO1 mRNA levels, showed (Fig. 1I). Surprisingly, the cancer cell line encyclopedia (CCLE) dataset revealed that H2228 cells also expressed elevated mRNA levels of IL-1α and CXCL8 (downstream target of IL-1 signaling) (Supplemental Fig. 1B). Given that IL-1α also induced similar signaling cascade as IL-1β, we next asked whether autocrine signaling by IL-1α in H2228 cells is responsible for elevated IDO1 mRNA levels in these cells. We incubated H2228 cells with IL-1 receptor antagonist (IL-1Ra), a natural antagonist to IL-1 signaling, and observed a decrease in IDO1 mRNA (Fig. 1I), indicating that autocrine signaling through IL-1α could be contributing to the elevated IDO1 levels in these cells.

A significant increase in IDO1 protein was observed in H1792 and HCC827 cells treated with IL-1β (Fig. 1J). We tested the effect of the potent inhibitor of IDO1 activity (IDOi) epacadostat [40] on monolayer cultures of H1792 and HCC827 cells treated with IL-1β. Cell counts revealed that epacadostat at 20 nM concentration for 4 days did not decrease viability of these cells (Fig. 1K), indicating that IDO1 activity may not directly induce lung cancer cell growth in monolayer cultures. Our data indicate that regardless of the driver mutations, lung adenocarcinoma cells induce immunosuppressive IDO1 mRNA and protein, but not TDO2, in response to IL-1β stimulation.

Culturing lung cancer cells in 3D enhances the induction of IDO1 mRNA and protein by IL-1β

To account for complexity of cell-to-cell contacts and formation of a hypoxic core formed under 3D growth conditions, we tested whether IL-1β–mediated induction of IDO1 in lung adenocarcinoma cells was conserved in a spheroid model of cell culture. We chose to study HCC827 and H1792 cell lines since they had the highest induction of IDO1 among the lung cancer cell lines studied (Fig. 1D) and they form spontaneous spheroids in low attachment conditions. Spheroids were formed in low attachment culture conditions in medium with or without 5 ng/mL IL-1β for 5 days. Given the spheroids take about 48 h to form, they were incubated for an additional 72 h before measuring the changes in IDO1 expression. Monolayer cultures were established for the same duration and conditions in parallel. Both cell lines displayed a more dramatic induction of IDO1 mRNA (Fig. 2A) and protein (Fig. 2B) when grown as spheroids compared to monolayer cultures, possibly due to the altered properties of the cells cultured in the 3D shape. Intriguingly, HCC827 cells appear to have higher IDO1 mRNA and protein in spheroid culture than in the monolayer cultures, which was also reported in CFPAC-1 pancreatic adenocarcinoma cells [41]. Similar induction was also observed in triple-negative breast cancer cells, but intriguingly, only TDO2, was basally higher in cells cultured under low attachment conditions than in monolayer cultures [42], indicating a cell-type or cancer-type specificity.

Fig. 2

Regulation of IL-1β-mediated IDO1 induction in 3D spheroid model of lung adenocarcinoma cells. A RT-qPCR measurement of IDO1 mRNA in H1792 and HCC827 monolayer cells and spheroids generated in 1% low attachment dishes, in control or IL-1β containing medium for 5 days. mRNA levels were normalized to 18s mRNA and represented as fold change over control treatment. B Western blot for IDO1 protein in H1792 and HCC827 monolayer and spheroid cultures ± 5 ng/mL IL-1β for 5 days. GAPDH was used as the loading control. C Viability of H1792 and HCC827 spheroids stimulated with 5 ng/mL IL-1β ± 20 nM Epacadostat (IDOi) for 5 days was measured with CellTiterGlo 3D assay (n = 4 spheroids). Luminescence was measured and viability is represented as relative luminescence units (RLU). D IL-1β induces IDO1 expression within the cells of lung adenocarcinoma spheroids. HCC827 spheroids were grown in low attachment conditions in vehicle control or 5 ng/ml IL-1β for 5 days were immunostained for IDO1 (Texas red, red) and nucleus (DAPI, blue). E IL-1β increases the percentage of IDO1+ cells in the spheroid. The number of IDO1 + cells were counted and represented as a percentage of the total cell population for n = 4 images. Each image of spheroid counted contained between 60—185 total cells. Error bars represent ± SD of 3 biological replicates; *P ≤ 0.05, ** ≤ 0.005, *** ≤ 0.0005

We next tested whether inhibiting IDO1 activity could suppress spheroid growth of lung adenocarcinoma cells. Surprisingly, IL-1β induced spheroid growth of HCC827 cells, but not H1792 cells, and this increased spheroid growth was modestly inhibited by epacadostat (Fig. 2C). Further, immunofluorescence revealed that a subset of HCC827 cells had elevated IDO1 protein when exposed to IL-1β (Fig. 2D). IDO1+ cells were present throughout the spheroid. About 42% IDO1+ cells were present within an IL-1β–treated spheroid and 8% in control-treated cells (Fig. 2E). Our data indicates that IL-1β-mediated induction of IDO1 is conserved in spheroid model of cell culture.

IL-1β regulates IDO1 transcription by activating the NFκB pathway in lung adenocarcinoma

An early event in IL-1 signaling is the activation of p65 (RelA gene) [43]. To determine whether p65 mediates IL-1β induction of IDO1, we silenced RelA mRNA in H1792 and HCC827 cells and then treated with vehicle or IL-1β (5 ng/mL) for 48 h. RT-qPCR and Western blotting showed that RelA mRNA (Fig. 3A) was effectively knocked down. To assess whether transcriptional activity of p65 was repressed, we measured the mRNA expression of CXCL8, a known gene target of the p65 transcription factor. CXCL8 mRNA (Fig. 3B) was also downregulated in both cell lines, indicating that p65 transcriptional activity was also reduced. We then analyzed the effect of p65 knockdown on IDO1 mRNA levels. RelA knockdown repressed the IL-1β mediated mRNA (Fig. 3C) and protein induction (Fig. 3D) in H1792 and HCC827 cells. Additionally, we used BAY-17085, an inhibitor of p65 nuclear translocation, to further validate the role of p65 in inducing IDO1 in adenocarcinoma cells (Fig. 3E). Our results demonstrated that the NFκB transcription factor p65 regulated IDO1 in IL-1β-stimulated cells.

Fig. 3

IL-1β regulates IDO1 transcription by activating the NFκB pathway in lung adenocarcinoma. A H1792 and HCC827 cells transfected with 40 nM of non-targeting or RELA/p65 siRNA (Dharmacon) for 24 h followed by treatment with vehicle (ctrl) or 5 ng/mL IL-1β for 48 h and RT-qPCR for RELA mRNA was performed to validate the efficacy of knockdown. B RT-qPCR for CXCL8 mRNA indicates repression of p65 transcriptional activity in H1792 and HCC827 cells. C RT-qPCR for IDO1 mRNA levels indicates p65 regulates IDO1 transcription in H1792 and HCC827 cells. D Western blot showing levels of IDO1 and p65/RelA protein levels. E Western blot of HCC827 cells treated with 5 ng/ml IL-1β ± BAY11-7085 (p65/NFκB inhibitor) for 48 h. Error bars represent ± SD of 3 biological replicates: *P ≤ 0.05, ** ≤ 0.005, *** ≤ 0.0005. mRNA levels were normalized to 18s mRNA and represented as fold change over control treatment. GAPDH is the Western blot loading control

IL-1β induces enzymatically active IDO1 protein

To quantify the enzymatic activity of IDO1 in IL-1β–treated cells, we first determined the kinetics of IDO1 induction by IL-1β using HCC827 cells. IDO1 mRNA and protein induction was maximal at 6 h after treatment, and the levels decreased until 24 h and then stayed constant until 48 h (Fig. 4A, B). We measured the intracellular and extracellular Kyn levels at 6, 12, 24 and 48 h using LC–MS. Mass spectrometry analysis revealed that intracellular Kyn levels increased significantly in the cells stimulated with IL-1β relative to the control by 6 h post treatment and a continued increase up to 48 h later (Fig. 4D); intracellular Trp levels remained unchanged. Extracellular Kyn levels were also detected in the medium by 6 h post IL-1β treatment relative to the control, and continued to increase to 48 h post treatment (Fig. 4E) and the extracellular Trp levels were significantly lower at 12 and 48 h. These data show that enzymatically active IDO1 protein is induced by IL-1β in lung adenocarcinoma cells.

Fig. 4

Measurement of IDO1 activity. A Level of IDO1 mRNA induction over time was measured by RT-qPCR of HCC827 cells grown in media with vehicle control or 5 ng/ml IL-1β for 6, 12, 24 or 48 h. mRNA levels were normalized to 18s mRNA and represented as fold change over control at 6 h. B Accumulation of IDO1 protein over time is shown by western blotting. GAPDH was used as the loading control. C Schematic of the Trp/Kyn pathway. D – E Intracellular (D) and extracellular (E) L-Trp and L-Kyn levels were measured as an indicator of IDO1 enzymatic activity. Relative metabolite content is normalized to total ion count. Error bars represent ± SD of 3 biological replicates: *P ≤ 0.05, ** ≤ 0.005, *** ≤ 0.0005

IL-1β is a common upstream regulator of immune checkpoint proteins

To gain better insights into the biological effects of IL-1β in lung cancer cells, we used the TCGA database to identify genes that are co-expressed with IDO1 in lung adenocarcinoma patients. We were intrigued to see that the PD-1 ligands PD-L1(CD274 gene) and PD-L2 (PDCD1LG2 gene) mRNA were among top significant genes that were positively correlated with IDO1 mRNA in lung adenocarcinoma tumors (Fig. 5A, B). To determine whether IL-1β is an upstream regulator of these immune checkpoint proteins, we measured PD-L1 and PD-L2 mRNA levels in A549, H1792, H2347, HCC364 and HCC827 in response to IL-1β treatment. Significant upregulation of CD274 (PD-L1) and PDCD1LG2 (PD-L2) mRNA (Fig. 5C, D) and PD-L1 protein (Fig. 5E) was detected. Several studies have previously reported that NFκB signaling mediates PD-L1 mRNA and protein induction in cancer cells [44]. NFκB-p65 signaling however, did not induce PD-L1 or PD-L2 mRNA or PD-L1 protein in IL-1β–stimulated HCC827 cells (Supplemental Fig. 2A, B, C). This indicates that IL-1β could be coactivating other pathways or transcription factors to induce PD-L1 and PD-L2 expression in lung adenocarcinoma cells.

Fig. 5

IL-1β mediates PD-L1 and PD-L2 induction concomitant with IDO1 induction in lung adenocarcinoma cells. A Expression of CD274 and IDO1 mRNA in human lung adenocarcinoma. Each dot represents a lung adenocarcinoma tumor from TCGA (n = 510). B Expression of PDCD1LG2 and IDO1 mRNA in human lung adenocarcinoma. Each dot represents a lung adenocarcinoma tumor from TCGA (n = 510). C—D RT-qPCR measurement of CD274 mRNA (C) and PDCD1LG2 mRNA (D) induction in lung adenocarcinoma cells stimulated with 5 ng/mL IL-1β for 48 h. E Western blots showing PD-L1 protein (CD274 gene) and IDO1 protein induction in lung adenocarcinoma cells. F Schematic showing the experimental set-up for (G). G Western blot for IDO1 and PD-L1 protein in HCC827 cells at 24 and 48 h after 5 ng/mL IL-1β stimulation and their levels 24 and 48 h after the removal of the stimulation. GAPDH was used as the loading control. Error bars represent ± SD of 3 biological replicates: *P ≤ 0.05, ** ≤ 0.005, *** ≤ 0.0005. mRNA levels were normalized to 18s mRNA and represented as fold change over control treatment

We tested whether the transcription factors that have been reported to induce PD-L1 in other cell or cancer types, including CEBPβ, NRF2 [45] and RelB [46] induced PD-L1. However, in IL-1β-stimulated lung adenocarcinoma cells, these transcription factors did not affect the expression of PD-L1 or PD-L2 (Supplemental Fig. 2). STAT signaling is one of the most well-characterized modes of PD-L1 induction in cancer cells [47,48,49], therefore we tested whether IL-1β induces STAT activation in lung cancer cells. Since IL-1β primarily signals though NFκB pathway [43], we were surprised to see an increase in p-STAT1 and p-STAT3 levels in HCC827 cells treated with IL-1β (Supplemental Fig. 2E).

To assess the duration of elevated protein levels following the cessation of cytokine stimulation, we exposed HCC827 cells to either a control or 5 ng/mL IL-1β treatment for 48 h. Subsequently, the cells were rinsed, and IL-1β-free medium was introduced (Fig. 5F). Both PD-L1 and IDO1 protein levels remain elevated 24 h after IL-1β was removed (Fig. 5G), and only IDO1 levels remained elevated 48 h later. Additionally, we see that the lung cancer cells are sensitive to the species that the cytokine is derived from. The human lung cancer cells require 100 times more of the mouse derived IL-1β to generate an IDO1/PD-L1 response that is seen with the human-derived IL-1β (Supplemental Fig. 3), indicating that further in vivo characterization needs to be performed in syngeneic or humanized mouse models. Together, these data show that IL-1β is indeed an upstream regulator of immune checkpoint proteins IDO1, PD-L1 and PD-L2.

Here, we show that IL-1β-stimulated lung adenocarcinoma cells with various driver mutations respond with an NFκB-mediated upregulation of IDO1. This IDO1 induction is concomitant with an increase in PD-L1 and PD-L2 and together, may provide the lung cancer cells a survival advantage by Kyn-dependent or PD-L1/2-PD-1-mediated inhibition of T-cell activity (Fig. 6).

Fig. 6

Working model for the regulation of immunomodulators by IL1β in NSCLC