The relationship between expression of PD-L1 and HIF-1α in glioma cells under hypoxia

1To the Editor

2In clinical, the blockade of the PD-1/PD-L1 pathway havn’t been well-confirmed to prolong OS of glioma patients. With increasing malignancy, hypoxia as a major tumor microenvironment factor widely exhibit in glioma, 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.

3PD-L1 is wildly used as a candidate biomarker for predicting patients that would respond to anti-PD-1/PD-L1 immunotherapy, but not in glioma. 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.

4Next, 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).

5To 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).

6To 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, 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.

7We 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.

8Our 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.