Wnt5b-induced PCP/JNK activation promotes PD-L1 expression and the malignant phenotype of non-small cell lung cancer

Background: Wnt5b is noncanonical Wnt ligand, and programmed-death ligand 1 (PD-L1) is a targeted agent for immunotherapy, but the mechanism by which Wnt5b regulates PD-L1 expression in non-small cell lung cancer (NSCLC) is unclear. Methods: Wnt5b and PD-L1 expressions were detected in NSCLC specimens by immunohistochemistry. The interrelationship connecting Wnt5b with PD-L1 was verified using dual-luciferase assay, immunofluorescence, coimmunoprecipitation, western blot ， real-time PCR and xenograft tumor model. Results: Wnt5b and PD-L1 expressions were positively correlated in NSCLC specimens. Five-year survival time in the group with their coexpression was significantly lower than that without coexpression. Under the effect of Wnt5b, Frizzled-3 (Fzd3) initiated Dishevellde-3 (Dvl-3) membrane recruitment via DEP domain by Dvl-3 phosphorylation, contributing to activate PCP/JNK signaling through the small GTPase Rac1, and then upregulate PD-L1 expression and promote the malignant phenotype of NSCLC in vivo and in vitro. After PD-L1 antibody treatment, Wnt5b induced tumor growth was inhibited significantly in xenograft tumor model. Conclusion: We demonstrate a new signal transduction pathway: Wnt5b initiates Dvl-3 membrane recruitment via DEP domain by Fzd3 so as to promote Rac1–PCP/JNK–PD-L1 pathway, which provides a potential target for clinical intervention and immunotherapy in lung cancer.

However, the regulatory mechanisms of PD-L1 expression by Wnt signaling in NSCLC has not been reported.
It binds to Wnt ligand and then interacts with Dvl to promote the Dvl phosphorylation for downstream cascade reaction [19,20].
In summarize, we speculate that Wnt5b may impact one or more domains of Dvl-3 via a specific Fzd isoform so as to active Wnt/PCP signaling and regulate PD-L1 expression in NSCLC. In this work, we examined the Wnt5b and PD-L1 expression in NSCLC specimens. NSCLC xenograft murine models induced by Wnt5b combination with an anti-mouse PD-L1 antibody were established to explore the efficacy of anti-PD-L1 therapy in NSCLC. And we aslo elucidated the mechanism by which Wnt5b regulates PD-L1 expression involved in Wnt/PCP pathway.

Patients and specimens
We collected 137 NSCLC specimens (average age: 60 years). From 2004 to 2014, these patients underwent surgery at the First Affiliated Hospital of China Medical University, and none of them received radiotherapy or chemotherapy before resection.  [22]. Complete follow-up data was available for all the patients. Clinicopathological information is in Table 1.
We aslo collected 22 fresh paired carcinoma and adjacent noncancerous tissues, which were immediately stored at −70°C for mRNA and protein extraction.
PD-L1 staining was performed by using the Dako automated staining platform according to the manufacturer's standard steps (Dako 22C3 PharmDx Assay, Dako Autostainer Link 48, Agilent, Santa Clara, CA, USA). We determined the PD-L1 expression in tumor cells by using Tumor Proportion Score (TPS). TPS was calculated as the percentage of at least 100 viable tumor cells with complete or partial membrane staining. The percentage of ≥50% was defined as PD-L1 positive expression; and that of＜50% was defined as PD-L1 positive expression.

RNA extraction and Real-time PCR
We used PrimeScript™ RT reagent Kit (TaKaRa, Dalian, China) to perform Real-time PCR according to manufacturer's instruction. The experimental steps are conducted as the previous study [13]. β-actin was normalized reference. The sequences of primers in this research were shown in Table S1.

Western blot and immunoprecipitation
The experimental steps are conducted as the previous study [13]. We separated the membrane and cytosolic proteins by using Membrane and Cytosol Protein Extraction kit according to manufacturer's instruction (Beyotime Institute of Biotechnology, Haimen, China). Detailed information of the antibodies is in Table S2.

Immunofluorescence staining
The experimental steps are conducted as the previous study [13]. After being treated with corresponding factors, the cell line was incubated overnight with an antibody against myc (1:50).

Dual-luciferase assay
We plated cells in 24-well plate. After incubation for 24 h, we transfected the mixture

Transwell assay
Transwell membranes (8 µm pore polycarbonate membrane, in 24-well plates) were used. We used Matrigel (BD Bioscience) on the upper surface of the membrane in every well. In the upper chamber, we cultured the cells (3 × 10 5 cells) in 100 µL serum-free medium; in the lower chamber, we added the medium containing 10% FBS. After incubation for 20 h, we stained the cells with hematoxylin (Sigma). Ten fields (400 × magnification) of each filter were randomly selected under microscope for counting the number of the invaded cells.

Cell scratch experiment
We seeded cells in six-well plates. After being treated with corresponding factors, the well was scratched by a 100 µL pipette tip. At 1 h and 24 h time points, we measured the scratch width under the microscope at three separated sites. The experiments were performed for three times.

Cdc42 and Rac1 GTPase Activation Assay
Rac1/cdc42 activation assay kit (Sigma-Aldrich, Deisenhofen, Germany) was used in this study according to manufacturer's instructions. After the pulldown reaction with GST-PAK-PBD (p21 binding domain of p21 activated kinase) beads, the precipitated proteins binding to the beads were performed to take immunoblotting assay with Rac1/cdc42 monoclonal antibody. We aslo determined the total amount of Rac1 or cdc42 by Western blot using total cell lysates.

Colony formation and MTT assays
The experimental steps are conducted as the previous study [13]. The experiments were performed for three times.

Transplantation of tumor cells into nude mice
The mice were treated in accordance with the experimental animal ethics guidelines  Wnt5b and PD-L1 coexpression was significantly shorter than that without 20 coexpression (P<0.05; Fig. 1e). S1b and S2a-d).

29
To assess the effect of Wnt5b on the tumorigenesis, NSCLC cells were intravenously  reporter assays with an ATF2 reporter system, which is a robust readout for JNK 46 signaling in Xenopus embryos [24]. Wnt5b overexpression strongly activated the 47 ATF2 luciferase reporter, whereas Wnt5b knockdown had the opposite effect (Fig. 3a).

130
And cotransfection GFP-Wnt5b with myc-Dvl-3 promoted Rac1 activation, but this 131 effect was abrogated upon the deletion of Dvl-3 DEP domain, as well as the deletion 132 of both the DEP and PDZ domains, but not the PDZ or DIX domain (Fig. 6b).

133
However, the effect of Wnt5b mediated cdc42 activation is not affected by any 134 domain deletion (Fig. S10c). The effect of Wnt5b mediated JNK activation was 135 significantly inhibited by adding Rac1 inhibitor, but this effect is not affected by 136 18 cdc42 inhibitor (Fig. 6c) by JNK inhibitor (Fig. 6d). It confirmed that JNK was important for Wnt5b induced 146 PD-L1 upregulation. 147 We also evaluated the effect of PD-L1 antibody in vivo. After anti-mouse-PD-L1 148 antibody treatment, Wnt5b induced tumorigenesis was inhibited significantly (A549: