Deubiquitination and Stabilization of PD-L1 by USP21

Background: The immunotherapy for different types of cancers that targeting programmed death protein-1 (PD-1) and programmed death ligand-1 (PD-L1) has highlighted the importance of suppressing specic T cell responses. Recently, several studies have shown that the expression level of PD-L1 in tumor cells is positively correlated with tumor metastasis as well as recurrence rate. The potent effects of post-translational modications (PTMs) for PD-L1, such as ubiquitination, glycosylation, phosphorylation and palmitoylation, have been reported to be related to immunosuppression. However, the regulation of PD-L1 degradation in cancers is still not well understood. In this paper, we mainly investigate the deubiquitination regulation of PD-L1. Methods: The protein levels of PD-L1 and USP21 were detected by Immunoblotting and immunohistochemistry. The interaction between PD-L1 and USP21 was determined by co-immunoprecipitation. The deubiquitination of PD-L1 was determined by in vitro deubiquitination assay. The deubiquitination sites of PD-L1 were identied by mass spectrometry analysis. The expression of mRNA in target tissues was presented by bioinformatics analysis. Results: Overexpression of USP21 signicantly increased PD-L1 abundance and knockdown of USP21 induced degradation of PD-L1. In vitro deubiquitination assay showed that USP21-WT reduced polyubiquitin chains from PD-L1 while USP21-C221A did not. Furthermore, ve lysines in intracellular segment of PD-L1 are potential deubiquitin sites and cancer-derived mutations of PD-L1 in Asp276 have the ability to enhance the deubiquitination of PD-L1 mediated by USP21. Finally, we found that USP21 is the frequently amplied deubiquitinase in lung cancer, especially in lung squamous cell carcinoma, and its amplication co-occurs with the upregulation of PD-L1 levels. Moreover, IHC analysis showed stronger staining of PD-L1 and USP21 in lung cancer samples than adjacent tissues. Conclusion: We identied USP21 as a novel deubiquitinase of PD-L1. Hopefully, targeting PD-L1 by inhibiting USP21 might be a potentially novel strategy for the treatment of lung cancer. carcinoma;


Introduction
Programmed cell death ligand-1 (PD-L1), also known as cluster of differentiation 274 (CD274) and B7 homolog 1 (B7-H1), is a transmembrane protein expressed on the surface of B cells, activated T cells, macrophages, dendritic cells and tumor cells (Huang et al. 2019). The binding of immune inhibitory receptor PD-1 (programmed death protein-1) with its ligands triggers the inhibition of effector T cells, results in cancer cells evasion from anti-tumor immunity (Petrelli et al. 2018).
In order to recover the response of cytotoxic T cells on tumors, a great many of monoclonal antibodies have been developed to prevent the interaction between PD-1 and PD-L1 (Huang et  In this study, we investigate the effects of USP21 on the deubiquitination and stabilization of PD-L1 and imply that targeting USP21 is a potential treatment for lung cancer.

Methods
Cell culture and transfection. Plasmids.
HEK293T cells were transfected with indicated plasmids and collected in RIPA buffer at 4 °C for 30 min.
After 15,000 g centrifugation at 4 °C, supernatant and following conjugated beads were combined and rotated at 4 °C for 6 h: anti-Flag (DYKDDDDK) a nity gel (#B23102, Bimake) and anti-HA magnetic (#B26202, Bimake) beads. The beads were collected and washed with RIPA buffer 4 times and heated at 100 °C for 10 min in 2 × SDS loading buffer prior to IB. HeLa cells were collected in RIPA buffer at 4 °C for 40 min. After 15,000 g centrifugation at 4 °C, supernatant from samples were pre-cleared to remove nonspeci c interactions with protein A/G beads rotating at 4 °C for 2 h. Following pre-clearing and PD-L1 antibody conjugation (4 °C for 2 h), lysate and beads were combined and rotated at 4 °C for 6 h.
Following incubation, A/G beads were collected and washed with RIPA buffer 4 times and heated at 100 °C for 10 min in 2 × SDS loading buffer prior to IB.
HEK293T cells were transfected with Flag-PD-L1, HA-USP21, HA-USP21-C221A expression plasmids. Cells were collected 48 h post transfection following 6 h MG-132 (20 µM) treatment. Cells were lysed using a RIPA lysis buffer with protease inhibitor and centrifuged at 15,000 g for 10 min. The supernatant containing protein was incubated with Flag/HA-beads for 6 h. PD-L1 and USP21 were puri ed from the cell extracts with Flag/HA-beads in lysis buffer. For deubiquitination assay, ubiquitinated PD-L1 was incubated with HA-USP21 or HA-USP21-C221A in DUB Buffer (50 mM Tris pH 7.5, 50 mM NaCl, 5 mM DTT, 5 mM MgCl 2 ). Reactions were incubated at 30 °C for 2 h on rotary shaker. Next, proteins were eluted from beads and heated at 100 °C. Ubiquitination status of PD-L1 was assessed by IB as described above.
Statistical analyses were performed using the two-tailed Student's t-test. Data were analyzed using the Prism software (GraphPad, San Diego, CA, USA) and expressed as the mean ± S.D. p < 0.05 was considered to be signi cant.

Results
USP21 binds and regulates the stability of PD-L1.
In our previous research, we have identi ed USP21 as a potential effector of PD-L1. To determine whether PD-L1 stability is positively controlled by USP21, we examined protein accumulation of PD-L1 in HEK293T and H1975 cells. Immunoblot analysis reveal that PD-L1 abundance signi cantly increased with overexpression of USP21 (Fig. 1a-b). In addition, knockdown of USP21 induced degradation of endogenous PD-L1 in HEK293T cells (Fig. 1c). These results indicate that PD-L1 abundance is regulated by USP21. Next, a robust endogenous interaction between PD-L1 and USP21 was detected in HeLa cells (Fig. 1d). Correspondingly, an exogenous interaction between PD-L1 and USP21 was also detected in HEK293T cells by co-immunoprecipitation (Fig. 1e). These results reveal that USP21 directly interacted with PD-L1. Furthermore, to determine the domain on PD-L1 recognized by USP21, we analyzed the interaction between USP21 and different truncations of PD-L1 (Fig. 1f). These results reveal that USP21 binds to PD-L1 through the intracellular (C-terminal 262-290 aa) and extracellular domains (N-terminal 19-241 aa) (Fig. 1g).
USP21 is a deubiquitinase that remove polyubiquitin chains from its substrates and has been shown to cleave K48 polyubiquitin linkages in vitro (Ye et al. 2011). Previous studies have shown that PD-L1 levels were regulated by K48-linked polyubiquitination and proteasomal degradation (Li et al. 2016). To determine whether USP21 affects the binding of polyubiquitin chains on PD-L1, we analyzed PD-L1 ubiquitination in the presence of MG132, a proteasome inhibitor. We found that MG132-induced PD-L1 ubiquitination was abolished by USP21 overexpression (Fig. 2a). However, the level of polyubiquitinated PD-L1 was restored when USP21-C221A (catalytically inactive mutant) was overexpressed, con rming the involvement of the catalytic activity of this deubiquitinase (Fig. 3a). Next, we performed an in vitro deubiquitination assay using USP21-WT and USP21-C221A immunopuri ed from HEK293T cells. Similarly, USP21-WT reduced polyubiquitin chains of PD-L1 (Fig. 2b), while USP21-C221A did not (Fig. 3b). These data suggest USP21 is the deubiquitinase of PD-L1.
PD-L1 is a transmembrane protein, including N-terminal signal peptide, extracellular domain, transmembrane domain and intracellular domain (Zhang, Bu, et al. 2018). To explore the mechanism behind USP21-induced PD-L1 stability, we detected whether PD-L1 undergoes any deubiquitination changes following co-incubating with USP21. As determined by mass spectrometry, USP21 induced several PD-L1 deubiquitination events (Fig. 2c). We found the deubiquitination sites are mainly located in the intracellular domain of PD-L1, such as K280, K281 and K270 (Fig. 2c). Thus, we hypothesized that USP21 deubiquitinates PD-L1 mainly through intracellular domain and we found ve potential sites (Fig. 2d). Mutating these residues to ubiquitination-resistant arginine showed that single-site mutant (K263R, K270R, K271R, K280R or K281R) had no sigini cant effect on the ubiquitination level of PD-L1 (Fig. 2e), while 5KR mutation almost abolished ubiquitination levels (Fig. 2f). These results indicate that these ve lysines are potential deubiquitin sites of PD-L1.
Cancer-derived mutations of PD-L1 disrupt the regulation of deubiquitination.
Catalogue of Somatic Mutations in Cancer (COSMIC) database reveal that PD-L1 mutations occur in 1.6% of human cancers and are largely clustered within the intracellular domain (Fig. 3c). To further explore the effect of cancer-derived mutations of PD-L1 in deubiquitination mediated by USP21, we constructed mutants in intracellular domain of PD-L1, we found these mutants (R260C, I274V, D276Y, D276H, T277S, T277K, T290M, K280N) had little effect on ubiquitination of PD-L1, while the two mutants in Asp276 (D276Y and D276H) enhanced deubiquitination levels (Fig. 3d-e). These results suggest that inhibition of USP21 may provide a better therapeutic strategy on cancers with PD-L1 Asp276 mutations.
PD-L1 and USP21 are up-regulated in lung cancer.
Finally, we analyzed the gene expression frequency of PD-L1 and USP21 in 17 types of cancers through The Cancer Genome Atlas (TCGA) database (Fig. 4a). We found the proportions of highly expressed PD-L1 and USP21 were higher in lung cancer and mRNA levels of PD-L1 and USP21 were overexpressed when compared to normal tissues (p < 0.001) (Fig. 4a-b). In addition, it should be noted that the mRNA levels of PD-L1 and USP21 were higher in lung squamous cell carcinoma (LUSC) than that in lung adenocarcinoma (LUAD) (Fig. 4c). USP21 ampli cation was also observed in other cancer types (Fig. 4a), which is consistent with previous studies that USP21 regulates cell proliferation and metastasis in colorectal cancer and pancreas cancer (Yun et al. 2020;Hou et al. 2019). Next, we con rmed the expression levels of PD-L1 and USP21 in LUSC and found that upregulated expression of PD-L1 accounted for 80% (395 of 494) of LUSC samples, while USP21 accounted for 76% (375 of 494) of LUSC samples (Fig. 4d). Furthermore, the expression frequency of USP21 in different clinical stages of lung cancer is consistent with that in PD-L1. (Fig. 4e-f). Together, these results suggest that USP21 is upregulated in lung cancer, especially in LUSC and its ampli cation co-occurs with upregulation of PD-L1.
Then we measured PD-L1 and USP21 levels in two paired lung cancer tumor samples and their adjacent normal tissues, IHC analysis showed stronger staining of PD-L1 and USP21 in lung cancer samples than adjacent tissues (Fig. 4g). In further research, we determined the deletion of last 28 amino acids of PD-L1 (262-290), impairs the interaction between PD-L1 and USP21, indicating that 262-290 region of PD-L1 is critical for the interaction with USP21. Notably, among the eight cancer-derived PD-L1 mutants located within intracellular domain, mutations at Asp276 enhanced the deubiquitination of PD-L1, which may lead to accumulation of PD-L1, allowing escape from anti-tumor immunity. These results suggest that the intracellular domain of PD-L1 is essential for the regulation of deubiquitination. USP21 has been shown to exhibit carcinogenesis and execute its carcinogenic function via its deubiquitination activity ). However, the role of USP21 in lung cancer remains unclear. In this paper, our results suggest that USP21 is up-regulated in lung cancer, especially in lung squamous cell carcinoma, the high expression rates of PD-L1 and USP21 both exceed 70%. Previous studies proved that CNS5 is the DUB of PD-L1 and the correlation between PD-L1 and CSN5 was con rmed in breast cancer (Lim et al. 2016). However, we found no correlation between PD-L1 and CSN5 in lung cancer. This interesting cancer-speci c difference in PTMs of PD-L1 makes the further research to explore how it regulates stabilization of PD-L1 and escape from anti-tumor immunity more valuable.

Conclusion
In conclusion, we identi ed USP21 as a novel deubiquitinase of PD-L1. USP21 induces PD-L1 stabilization and cancer escape from immunity via its deubiquitination levels. Importantly, USP21 is upregulated in lung cancer and co-occurs with the upregulation of PD-L1. Hopefully, targeting PD-L1 by inhibiting USP21 might be a potentially novel strategy for the treatment of lung cancer.

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

Ethics approval and consent to participate
The research was approved by the Ethics Committee of Zhejiang University. All patients provided written informed consent before this study.

Consent for publication
All authors consent for publications.