FGFR1 Promotes Tumor Immune Evasion via YAP-Mediated PD-L1 Expression Upregulation in Lung Squamous Cell Carcinoma

Background: Variations in fibroblast growth factor receptor 1 (FGFR1), which occur 2 frequently, are common driver mutations of lung squamous cell carcinoma. Immune 3 checkpoint inhibitors targeting programmed death-1 (PD-1) and programmed death 4 ligand-1 (PD-L1) are powerful anticancer weapons. Activation of FGFR1 leads to 5 tumorigenesis through multiple downstream molecules, including Yes-associated 6 protein (YAP), but whether and how FGFR1 regulates tumor immune evasion remain 7 largely unclear. 8 Methods: H520 and HCC95 cells were treated with siRNA and plasmids to increase or 9 decrease the expression of FGFR1, YAP and PD-L1, as assessed by molecular assays 10 of protein and mRNA expression. The interaction between YAP and PD-L1 was verified 11 by chromatin immunoprecipitation. After FGFR1 knockdown by shRNA, cancer cells 12 were cocultured with Jurkat T cells, and then cell proliferation and activity were 13 assessed. In C57BL/6 mice, the tumor immune microenvironment was analyzed by 14 flow cytometry, immunofluorescence and immunohistochemistry after FGFR1 15 knockdown. The effect of the combination of FGFR1 knockdown and PD-1 blockade 16 was explored both in vitro and in vivo. 17 Results: In H520 and HCC95 cells, FGFR1 upregulated PD-L1 expression via YAP, 18 and YAP initiated the transcription of PD-L1 after binding to its promoter region. Both 19 in vitro and in vivo, FGFR1 knockdown decreased tumor growth and reduced immune 20 escape and reactivation of T cells. The combination of FGFR1 knockdown and PD-1 21 blockade synergistically exerted antitumor effects. In human LSQCC, the expression 22 of fibroblast growth factor 2 (FGF2), the activator of FGFR1, was positively correlated 23 with that of PD-L1 at the mRNA level. 24 Conclusions: The FGFR1/YAP/PD-L1 regulatory axis mediates tumor-associated 25 immune suppression in lung squamous cell carcinoma, and FGFR1 knockdown 26 reactivates T cells in the tumor microenvironment. Synergistic inhibition of both 27 FGFR1 and PD-1/PD-L1 may be a possible treatment for lung cancer patients.


12
PD-L1 is abundantly expressed on cancer cells, macrophages, and dendritic cells (DCs). 13 The interaction between PD-1 and PD-L1 causes T lymphocyte dysfunction, enabling   increasing the immune escape of cancer cells 24-28 . We wondered whether YAP also 25 regulates the expression of PD-L1 in LSQCC. 26 In this study, we investigated the role of FGFR1 in the mechanisms of immune 27 escape by LSQCC cells and, more specifically, on immune cells of the TME. 28 Considering the limited response rate to targeted or immune therapy alone, the effect 29 of the combination of FGFR1 inhibition and PD-1/PD-L1 blockade was explored both 30 in vitro and in vivo. Our work uncovered a regulatory axis between FGFR1 and PD-L1 31 that may provide new insights into tumor immune evasion and clues for novel 32 therapeutic interventions for LSQCC patients. 33 34 Materials and methods 35 Cell culture and reagents 36 H520, HCC95 and Jurkat T cells were purchased from ATCC and cultured using 37 RPMI 1640 (HyClone, USA) with 10% fetal bovine serum (Gibco, USA), 100 U/mL 38 penicillin, and 100 μg/mL streptomycin (Gibco, USA). LLC cells were purchased 39 from ATCC and cultured using DMEM (HyClone, USA) with 10% fetal bovine serum 40 (Gibco, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco, USA).

41
FGF2 and IFN-γ were purchased from PeproTech. AZD4547 and nivolumab were 42 purchased from Selleck. AZD4547 was dissolved in DMSO, aliquoted, and stored as a 43 10 mM stock solution at −20 °C. Nivolumab was stored at 4 °C. Verteporfin (Sigma-Aldrich, USA) was stored at a concentration of 10 mM in DMSO at −20 °C. All cells 1 were maintained in a humidified incubator at 37°C with 5% carbon dioxide. Western blot analysis, RNA extraction, cDNA synthesis, and qPCR 11 Cells were lysed for protein or RNA extraction, and the samples were subjected to 12 western blot analysis or used for cDNA synthesis and qPCR as previously described 13 (5). The primers used for qPCR were as follows: Cell lysates were eluted with SDS-loading buffer (50 mM Tris-HCl, pH 6.8, 10% 32 glycerol, 1% SDS, 1% beta-mercaptoethanol). The eluates were separated by SDS- 33 PAGE and transferred to PVDF membranes (EMD Millipore, Germany). the  Western blotting images were captured with a ChemiScope5600 instrument (Clinx, 40 Shanghai, China) using ECL substrate.

42
Dual luciferase reporter assay 43 H520 cells were seeded in a 24-well plate until the cells reached approximately 50-60% confluence. For the luciferase assay, the cells were transiently transfected with the 1 indicated plasmids, and Renilla and firefly luciferase were used as internal controls.

2
The cells were lysed with passive lysis buffer (Promega, USA) 48 hours after 3 transfection and the indicated treatment. Luciferase assays were performed using a dual 4 luciferase assay kit (Promega, USA), and luciferase activity was normalized to that of 5 the internal control Renilla.

Coculture of cancer cells and T cells 23
Jurkat T cells were activated with 100 UI/ml IL-2 3 days before the experiment.

24
Cancer cells were inoculated in 24-well plates or 96-well plates 2 days in advance.

25
Complete culture medium containing 500 IU/ml IFN-γ was added to H520 and 26 HCC95 cancer cells for 24 hours to induce PD-L1 expression. Suspended/adherent 27 cells were added to the coculture system at a ratio of 2:1 to 8:1 and cultivated for 48-       Linear regression and Spearman correlation analysis of the mRNA levels of FGF2 and 1 PD-L1 was conducted as previously reported (5). The log-rank test was used for 2 statistical analysis. Single-cell RNA-seq gene expression data for tumor tissues from 3 NSCLC patients were downloaded from Gene Expression Omnibus (GEO). The 4 accession number of the dataset was GSE140819 (33). Seurat 4.0.1 was used to 5 standardize, integrate, cluster, visualize, and annotate the original data using an R tool 6 kit for single-cell genomics (33,34). were considered significant (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). 14 Except where otherwise indicated, the experiments were repeated three times.

15
Quantitative data are presented as the mean ± SEM. All images shown are 16 representative.

20
In H520 and HCC95 cells, protein and mRNA expression was measured after 21 treatment with small interfering RNAs (siFGFR1-1, siFGFR1-2 or siCTRL; 22 constructed and verified as previously reported (5)), an FGFR1/2/3 inhibitor (AZD4547) 23 and an FGFR1 activator (bFGF2). siFGFR1 and AZD4547 inhibited FGFR1, pFGFR1 24 and PD-L1 expression ( Figure 1A and B). There was no difference in LATS1 expression, 25 but the phosphorylation of LATS1 at two amino acid residues (S909 and T1079) was 26 decreased after FGFR1 inhibition ( Figure 1A and B). As demonstrated previously, a 27 decrease in phospho-LATS1 expression facilitated the phosphorylation of YAP at the 28 S127 residue. When trapped in the cytoplasm, YAP was inactivated through 29 phosphorylation ( Figure 1A and B). The total mRNA levels and protein levels of later, and it was found to be increased in the YAP-OE group (p < 0.001) ( Figure 2B).

41
This indicated that YAP acted on the promoter region of PD-L1. To confirm this, H520 42 cells were cultured with IFN-γ to stimulate the expression of PD-L1 and then lysed, and 43 DNA was extracted. Subsequently, chromatin immunoprecipitation was performed to analyze the cDNA library. The chromatin region precipitated by the YAP antibody was 1 located between 5,440,000 and 5,460,000 bases in the PD-L1 region of chromosome 9.

32
We further investigated the distribution of these molecules in the TME due to Ethics approval and consent to participate 20 All animal experiments were carried out in full accordance with protocols approved by 21 the Institutional Ethics Committee of Shanghai Jiao Tong University.

23
Consent for publication 24 Consent to publish has been obtained from all authors.

26
Availability of supporting data 27 The datasets used and/or analyzed during the current study are available from the 28 corresponding author on reasonable request.

30
Competing interests 31 The authors declare no potential conflicts of interest.  All authors read and approved the final manuscript.

9
Acknowledgements 10 We sincerely thank Haizhen Jin for the flow cytometry testing and guidance.