To the Editor

In clinical, the blockade of the PD-1/PD-L1 pathway havn’t been well-confirmed to prolong OS of glioma patients [1, 2]. With increasing malignancy, hypoxia as a major tumor microenvironment factor widely exhibit in glioma [3], however, the influence of hypoxia on tumor immune escape remains unclearly. Here, we aimed to explore the relationship between the PD-L1 and HIF-1α in glioma, and to investigate their prognostic values.

PD-L1 is wildly used as a candidate biomarker for predicting patients that would respond to anti-PD-1/PD-L1 immunotherapy [4, 5], but not in glioma [6]. We analyzed RNA-seq data from a cohort (640 glioma patients) in CGGA dataset and found that PD-L1 is positively correlated with HIF-1α (Additional file 1: Fig. S1, Additional file 2: S2). To determine this, the PD-L1 and HIF-1α levels in 120 glioma patients’ tissues were detected by immunohistochemical (Fig. 1a, Additional file 4: Table S1). Fifty patients (41.7%) were classified as PD-L1 positive (≥ 5%). PD-L1 was positively associated with tumor grade (Fig. 1b, Additional file 5: Table S2). Moreover, our clinical data showed that high PD-L1 was significantly related to high HIF-1α (r = 0.412, P < 0.001) (Fig. 1b). These findings were consistent with the external central nervous system tumors [7, 8].

Fig. 1
figure 1

The relationship of PD-L1 and HIF-1α expression in tumor tissue of glioma patients and their impact on the overall survival. a Immunohistochemistry (IHC) analysis of HIF-1α and PD-L1 in tissue sections of glioma patients. Typical image of positive expressions of HIF-1α (≥ 1%) and PD-L1(≥ 5%) in tissue sections of one patient with grade IV glioma; Typical image of negative expressions of HIF-1α (< 1%) and PD-L1 (< 5%) in tissue sections of one patient with grade II glioma. b PD-L1 and HIF-1α expression in patients with different grades glioma. PD-L1 and HIF-1α expressions in high-grade glioma (HGG) group and low-grade glioma (LGG) group; PD-L1 and HIF-1α expressions in grade II to grade III groups; Correlation analysis of PD-L1 and HIF-1α expression (r = 0.412, P < 0.001) in all glioma patients in our cohort. For (A) to (B), the data were presented as mean ± SEM. *P < 0.05, ***P < 0.001. c The overall survival of glioma patients. Statistical significance was determined by log-rank (Mantel-Cox) test

Full size image

Next, we investigated the correlation of PD-L1/HIF-1α expression and the OS of these glioma patients. The OS in either PD-L1 or HIF-1α positive group was significantly poorer than that in negative group (Fig. 1c). Subsequently, we classified all the patients with combining PD-L1 and HIF-1α expression into four subgroups. The Kaplan–Meier curves indicated that the patients in PD-L1(+) HIF-1α (+) group had worse OS than those in PD-L1(−) HIF-1α (−) group (P < 0.0001) (Fig. 1c). Univariate analysis identified PD-L1 ≥ 5%, HIF-1α ≥ 1%, HGG, and older age as unfavorable prognostic predictors (Additional file 6: Table S3). Multivariate analysis was also performed and indicated that both PD-L1 and HIF-1α expression were independent poor prognostic factors (Additional file 6: Table S3). We also found the consistent results in primary glioma patients in CGGA dataset (Additional file 2: Fig. S2).

To further verify the relationship between PD-L1 and HIF-1α, we cultured U251 and U343 glioma cell lines under hypoxic condition for different time and detected the PD-L1 expression. The western blot results showed higher HIF-1α and PD-L1 levels in hypoxia condition (1% O2) for either 24-, 48-, or 72-h culturing than those in control condition (21% O2) (Additional file 3: Fig. S3). Given these data, we used 1% O2 (72 h) as the hypoxia condition for further experiments. Similarly, we observed that PD-L1 expression was also increased with the hypoxia mimic CoCl2 treatment (Additional file 3: Fig. S3).

To further dissect the roles of the HIF-1α in PD-L1 up-regulation under hypoxia, we first knocked down HIF-1α using siRNA or inhibited HIF-1α activity using HIF-1α inhibitor (PX-478) and then detected PD-L1 expression. The results showed that either HIF-1α knockdown or PX-478 treatment can significantly decrease PD-L1 expression in glioma cells under hypoxia (Fig. 2a, b, Additional file 7: Table S4 and Additional file 8: Table S5). Given that HIF-1α protein can activate its target genes via directly binding to their promoter [8, 9], we verified whether PD-L1 is a direct target of HIF-1α in glioma cells using ChIP-qPCR assay. The results showed that HIF-1α directly interacts with the PD-L1 promoter region (~ 0.5 kb proximal to the transcription start site) (Fig. 2c and Additional file 9: Table S6). Furthermore, the co-staining PD-L1 and HIF-1α in glioma murine model showed that PD-L1 was highly express in hypoxic regions of tumors (Fig. 2d). These suggest that hypoxia upregulated PD-L1 via increasing HIF-1α in glioma cells.

Fig. 2
figure 2

Hypoxia up-regulate PD-L1 expression via HIF-1α in glioma cell lines and combination treatment with HIF-1α inhibitor and anti–PD-L1 antibody can reduce tumor growth in murine model of glioma. a qPCR analysis of HIF-1α and PD-L1 mRNA expression in U251 and U343 lines with different treatments as indicated. The qPCR data were normalized to GAPDH. The data were presented as mean ± SEM. P values were calculated by unpaired two-tailed Student’s t tests. *P < 0.05, **P < 0.01. b Western blot analysis of U251 and U343 cells with different treatments using indicated antibodies. c Chromatin immunoprecipitation (ChIP) analysis of the PD-L1 promoter in U251 cells using anti-HIF-1α mAb. The experiments were performed in triplicates and repeated three times. d Immunofluorescence staining of HIF-1α and PD-L1 expression in tumor cells analyzed by confocal microscopy. Representative images are shown. Scale bars, 50 μm. e Mice bearing GL261 cells were divided into the indicated treatment groups. The tumor volumes of mice treated with control, anti–PD-L1 monoclonal antibody, HIF-1α inhibitor (PX-478), or combined anti–PD-L1 antibody and PX-478 were measured and plotted (n = 5). Tumor volume was measured twice weekly. Data are presented as mean ± SEM. and the statistical significance was determined by two-way ANOVA. f Survival from mice receiving the indicated treatments as described in e. Statistical significance was determined by log-rank (Mantel-Cox) test. For (e) to (f) *P < 0.05, **P < 0.01. g The HE staining of intracranial tumor and immunohistochemistry analysis of CD8+ T cells in intracranial tumor from mice receiving control, anti–PD-L1 antibody, PX-478, or anti–PD-L1 antibody and PX-478. h Representative flow cytometry analysis and quantification of CD4+ T, CD8+ T, CD11c+ DC and CD11b+ myeloid cells populations in GL261 tumors with the indicated treatments (n = 5). i Quantification flow cytometry analysis of the PD-L1 expression on CD45−, CD3+, CD11c+ and CD11b+ cells (n = 5). j Representative flow cytometry analysis and quantification of CD8+ INF- γ+ T cells in GL261 tumors and the MFI of INF- γ in CD8+ T cells in U261 tumors at day 14 after treatment (n = 5). For (h) to (j), data are presented as means ± SEM. P values were calculated by unpaired two-tailed Student’s t tests. *P < 0.05, **P < 0.01

Full size image

We hypothesized that combining anti–PD-L1 and HIF-1α inhibitor would trigger an antitumor effect. Thus, we inoculated GL261 cells into wild type mice and treated the mice with anti–PD-L1 antibody and/or HIF-1α inhibitor. The combination treatment exerts a more pronounced antitumor effect, assessed in terms of both tumor growth and survival, than each monotherapy (Fig. 2e, f). Of interest was that, in situ glioma model (Luci+GL261), PX-478 can also enhance the intracranial efficacy of anti-PD-L1 antibody (Fig. 2g). Immunologically, our FACS results showed that the combination treatment significantly increased the percentage of tumor-infiltrated CD4+ T, CD8+ T, CD11c+ DC (Fig. 2h) and also decreased PD-L1 expression (Fig. 2i). Moreover, we also found the increased numbers of cytotoxic CD8+ T cells (IFNγ+CD8+) (Fig. 2j). Collectively, these indicate that combination treatment can reverse the immunosuppression microenvironment in glioma.

Our study demonstrated that positive relationship between HIF-1α and PD-L1 in glioma and provide the evidence that targeting HIF-1α can boost anti-PD-1/PD-L1 efficacy for glioma treatment.