Discovery of an OTUD3 inhibitor for the treatment of non-small cell lung cancer

The ubiquitin-proteasome system (UPS) controls protein turnover, and its dysfunction contributes to human diseases including cancer. Deubiquitinating enzymes (DUBs) remove ubiquitin from proteins to maintain their stability. Inhibition of DUBs could induce the degradation of selected oncoproteins and has therefore become a potential therapeutic strategy for cancer. The deubiquitylase OTUD3 was reported to promote lung tumorigenesis by stabilizing oncoprotein GRP78, implying that inhibition of OTUD3 may be a therapeutic strategy for lung cancer. Here, we report a small-molecule inhibitor of OTUD3 (named OTUDin3) by computer-aided virtual screening and biological experimental verification. OTUDin3 exhibited pronounced antiproliferative and proapoptotic effects by inhibiting deubiquitinating activity of OTUD3 in non-small-cell lung cancer (NSCLC) cell lines. Moreover, OTUDin3 efficaciously inhibited growth of lung cancer xenografts in mice. In summary, our results support OTUDin3 as a potent inhibitor of OTUD3, the inhibition of which may be a promising therapeutic strategy for NSCLC.


Introduction
The ubiquitin proteasome system (UPS) is essential for life in eukaryotes and regulates many aspects of cell physiology [1,2]. Ubiquitin (Ub) is covalently conjugated to protein substrates through the concerted action of activating (E1), conjugating (E2), and ligase (E3) enzymes and is removed by deubiquitinating enzymes (DUBs) [3]. Dysfunction of the ubiquitin proteasome system (UPS) has been implicated in the etiology and progression of a diverse set of human diseases including cancer, metabolic syndromes, neurodegeneration, autoimmunity, in ammatory disorders, and infection [4][5][6][7][8][9]. The UPS plays fundamental roles in the regulation of protein homeostasis and functions that are often altered during cancer progression, thus providing entry points for the development of anti-tumor therapies. The approval and clinical success of the proteasome inhibitor Velcade (bortezomib) and its successors has validated the UPS as a viable target for therapeutic intervention [10]. As part of the UPS pathway upstream of the proteasome, DUBs may offer new opportunities for the development of novel therapeutics with potentially enhanced speci city and reduced toxicity [11]. DUB inhibition increases ubiquitination of speci c substrates without affecting global protein or ubiquitin levels, which is among the most exciting and clinically applicable use of DUB inhibitors [12].
To date, no small-molecule inhibitors of OTUD3 have been reported. In this study, we screened 100 000 compounds available at ChemDiv for small-molecule inhibitors of OTUD3 by structure-based virtual screening. From this, we identi ed several primary hits that were further validated for inhibition of lung cancer cell growth by proliferation assays. OTUDin3, one of the candidate compounds, was identi ed using cell proliferation assays in vitro. Then, we con rmed that OTUDin3 bound to OTUD3 and inhibited the deubiquitinating activity of OTUD3 by interfering with Ub binding to OTUD3. Next, we demonstrated that OTUDin3 inhibited non-small-cell lung cancer (NSCLC) cell lines by enhancing the degradation of GRP78 both in vitro and in vivo. Collectively, we identi ed the rst small-molecule inhibitor of OTUD3, which effectively inhibited NSCLC. Our ndings support inhibition of OTUD3 as a promising therapeutic strategy to treat NSCLC.

Results
Identi cation of small-molecule OTUD3 inhibitors Previously, discovery of novel targeted inhibitors for DUBs has been mainly performed using highthroughput screening (HTS) of small molecule libraries [12]. Compared with traditional drug discovery methods, computer-aided drug design (CADD) is more e cient and economical. CADD integrates molecular docking to the ligand-binding pocket of a promising therapeutic target, computes the binding energy of each docked small-molecule compound, and selectively chooses the best ones as candidates for subsequent experimental procedures. Therefore, in this study we employed computational virtual screening based on molecular docking for the identi cation of OTUD3 inhibitors.
OTUD3 contains an ovarian tumor (OTU) domain and a ubiquitin-associated (UBA) domain (Fig. 1A). The OTU domain is the catalytic domain and is essential for OTUD3 to perform deubiquitinating activity and interact with substrates [23]. The OTUD3 (PDB: 4BOU) PDB le was downloaded from the Protein Data Bank (Fig. 1B). All of the heterogeneous atoms were removed, and the 4BOU chain A was selected for subsequent molecular docking. Previously developed inhibitors of DUBs have been found to bind the evolutionarily conserved catalytic centers of DUBs, resulting in poor selectivity [37]. However, a few inhibitors, such as the USP14 inhibitors IU1-series and USP7 inhibitor FT671, blocked access of the Cterminus of ubiquitin to the active site and showed surprisingly excellent selectivity [21,22,38]. Therefore, the S1 Ub-binding site was determined as the ligand binding site, which is key for the deubiquitinating activity of OTUD3 [39]. However, the structure of OTUD3 in complex with Ub was not yet resolved.
Considering that the structure of OTUD3 is highly similar to OTUD5 with a low root mean square deviation (~ 0.8 Å), we performed superimposition of the OTUD3 OTU structure with the structure of OTUD5 OTU-Ub complex to identify the Ub-binding site of OTUD3 (Fig. 1C, Supplementary Fig. 1A). Then, the docking GridBox was set to enclose the entire S1 Ub-binding site and all the amino acid residues around it (Fig. 1D). AutoDock Vina 1.1.2 was used for the subsequent molecular docking [40]. A diverse set of 100 000 compounds available at ChemDiv was screened, and 18 candidate hits with the highest docking scores were identi ed for further biological experimental veri cation (Fig. 1E, Supplementary table 1).
As OTUD3 plays a promoting role in lung tumorigenesis by stabilizing GRP78, inhibitors of OTUD3 should inhibit the proliferation of NSCLC cells [26]. Therefore, cell proliferation assays were chosen as a further screening strategy for the OTUD3 inhibitors. From this work, compound 16 was identi ed, which signi cantly inhibited the proliferation of H1299 cells after incubation with 10 µM of the candidate compound for 24 and 48 h (Fig. 1F, Supplementary Fig. 1B). The results were con rmed in H460 and A549 cell lines (Fig. 1F, Supplementary Fig. 1C).
Molecular docking analysis showed that the a nity energy of OTUDin3 to OTUD3 was − 10.1 kcal/mol, and OTUDin3 was docked into the S1 Ub-binding site pocket (Fig. 1H). The predicted protein-ligand interactions mode showed that OTUDin3 formed multiple interactions with OTUD3, including hydrophobic interactions with Asn136 and Ile162, hydrogen bonds with Gln154, Trp160, Tyr177, and Tyr183, and πstacking bonds with Trp160 and Tyr177 (Fig. 1I, Supplementary Fig. 1D).
To further con rm that OTUDin3 is an inhibitor of OTUD3, we rst performed surface plasma resonance (SRP) assays. The results showed that OTUDin3 bound to OTUD3 with a dissociation equilibrium constant (KD) value of 0.32 µM (Fig. 1J). To explore the speci city of OTUDin3, OTUD5 was selected to verify whether OTUDin3 binds to other DUB. The results of SPR showed that OTUDin3 did not bind to OTUD5 ( Supplementary Fig. 1E). This implied that OTUDin3 bound to OTUD3 with a certain speci city.
Next, we performed in vitro deubiquitination assays to verify whether OTUDin3 inhibits the deubiquitinating activity of OTUD3. The results indicated that OTUDin3 inhibited the deubiquitinating activity of OTUD3 to cleave K48-and K63-linked diUb (Fig. 1K, Supplementary Fig. 1F). Then, we performed in vitro OTUD3-Ub binding assays with OTUDin3 to verify whether OTUDin3 interfered with OTUD3 binding to Ub as designed. The results showed that OTUDin3 inhibited the binding of OTUD3 to Ub (Fig. 1L).
Collectively, we identi ed OTUDin3 as an inhibitor of OTUD3 using virtual screening and biological experiments, which represented a potentially tractable starting point suitable for further investigation.
OTUDin3 inhibits NSCLC cell growth, migration and invasion From preliminary screening, we found that OTUDin3 inhibited NSCLC cells at a concentration of 10 µM. Next, we used Cell Counting Kit 8 (CCK-8) assays to determine the 50% inhibitory concentration (IC 50 ) of OTUDin3. The results showed that the IC 50 value of OTUDin3 in H1299, A549, H460, and H1650 cells was 8.1, 14.8, 6.1, and 7.4 µM, respectively ( Fig. 2A, Supplementary Fig. 2A). Microscopic images of the cell morphology showed that OTUDin3 induced cell death in H1299 cells ( Supplementary Fig. 2B). Next, we detected the effects of OTUDin3 on normal cells, human pulmonary alveolar epithelial cells (HPAEpiC).
The results showed that OTUDin3 had less inhibition on HPAEpiC cells, and the IC 50 value was 102.3 µM which is an order of magnitude compared with lung cancer cells ( Supplementary Fig. 2C). In addition, colony formation assays showed that OTUDin3 inhibited H1299 and A549 cell growth in a concentrationdependent manner (Fig. 2B, Supplementary Fig. 2D). To con rm whether OTUDin3 inhibited the migration of lung cancer cells, we performed wound healing assays. The results showed that treatment with OTUDin3 signi cantly decreased the migration ability of H1299 cells in a concentration-dependent manner (Fig. 2C). Furthermore, OTUDin3 impaired the invasive ability of H1299 cells in a concentrationdependent manner, as assessed by Transwell assays (Fig. 2D). Collectively, these results demonstrated that OTUDin3 inhibited NSCLC cell growth, migration and invasion.

OTUDin3 induces NSCLC cell apoptosis
We next evaluated the ability of OTUDin3 to induce apoptosis. H1299 cells were treated with increasing concentrations of OTUDin3. The apoptotic cell population, including early and late apoptotic cells, was quanti ed using an apoptosis kit and analyzed by ow cytometry. The results showed that OTUDin3 induced apoptosis in H1299 cells in a concentration-dependent manner (Fig. 2E, F). Cell lysates were assessed by western blotting. Consistently, OTUDin3 increased the levels of apoptosis-associated proteins, Cleaved poly (ADP-ribose) polymerase (PARP) and Cleaved Caspase-3, in a concentrationdependent manner (Fig. 2G). Together, the results indicated that OTUDin3 induced NSCLC cell apoptosis.

OTUDin3 inhibits NSCLC cell growth, migration and invasion, and induces apoptosis by targeting OTUD3
To investigate whether the effects on NSCLC cells were mainly derived from targeting OTUD3, we knocked down OTUD3 in H1299 cells (Fig. 3A). Then, using the OTUD3 knockdown cells, we performed the same series of experiments previously performed. The CCK-8 assay results showed that compared with the control, the inhibitory effect on cell proliferation was signi cantly reduced in OTUD3 knockdown cells (Fig. 3B). Similarly, OTUDin3 had a weaker impact on colony formation in OTUD3 knockdown cells (Fig. 3C). Wound healing and Transwell assays showed that OTUD3 knockdown cells were less sensitive to OTUDin3 compared with the control (Fig. 3D, E). The results of apoptosis assay showed that OTUD3 knockdown cells were less responsive to OTUDin3, which was consistent with the above results (Fig. 3F, Supplementary Fig. 3A). However, we observed that OTUDin3 still inhibited cell growth after OTUD3 knockdown. We speculated that it may be due to low knockdown e ciency. Therefore, we constructed OTUD3 knockout (KO) cell line using the CRISPR/Cas9 technology ( Supplementary Fig. 3B), and then performed CCK-8 assays. The results showed that OTUDin3 had less inhibitory effect on the OTUD3 KO cells than on the OTUD3 knockdown cells ( Supplementary Fig. 3C). While OTUDin3 inhibited about 50% of OTUD3 wild-type (WT) cells at a concentration of 10 µM, it had little inhibitory effect on OTUD3 KO cells ( Supplementary Fig. 3C). Taken together, these results demonstrated that OTUDin3 inhibited lung cancer cell growth, migration and invasion, and induced apoptosis by targeting OTUD3. OTUDin3 downregulates GRP78 by inhibiting the deubiquitinating activity of OTUD3 We then investigated the underlying mechanisms by which OTUDin3 inhibited NSCLC cells. A previous study demonstrated that OTUD3 promotes lung tumorigenesis by stabilizing GRP78 [26]. Therefore, we performed a time-course analysis using western blotting to check the protein level of GRP78. H1299 cells were treated with 5 µM of OTUDin3 for the indicated time and then lysed. Western blotting analysis showed that OTUDin3 downregulated GRP78 in a time-dependent manner, and the downregulation effect was signi cant after 24 h (Fig. 4A). Then, H1299 cells were treated with increasing concentrations of OTUDin3 for 24 h. Western blotting analysis showed that OTUDin3 downregulated GRP78 in a concentration-dependent manner without affecting the stability of OTUD3 (Fig. 4B). The result was consistent with the downregulation of GRP78 by OTUD3 knockdown in the H1299 cell line[26].
OTUD3 upregulates GRP78 in a manner dependent on its deubiquitinating activity, as the catalytically inactive mutant C76A lost its ability to upregulate GRP78 [26]. To verify the downregulation effect of OTUDin3 on GRP78 resulting from inhibition of the deubiquitinating activity of OTUD3, GRP78 ubiquitylation assays were performed. We rst performed the ubiquitination assay of exogenous GRP78 in HEK293T cells. Flag-OTUD3 WT or Flag-OTUD3 C76A , Myc-GRP78, and HA-Ub were transfected into HEK293T cells as indicated. The different groups were treated with 5 µM of OTUDin3 or an equal volume of dimethyl sulfoxide (DMSO) for 46 h. The results showed that overexpression of OTUD3 WT signi cantly reduced the level of GRP78 ubiquitination, and the catalytically inactive mutant OTUD3 C76A did not reduce the level of GRP78 ubiquitination. However, when OTUD3 WT overexpressing cells were treated with OTUDin3, GRP78 ubiquitination was increased, reversing the deubiquitination of OTUD3 (Fig. 4C). The results indicated that OTUDin3 increased GRP78 ubiquitylation by inhibiting the deubiquitinating activity of OTUD3. Then, endogenous GRP78 ubiquitination assays were performed in H1299 cells to further con rm the above mechanism. The results showed that OTUDin3 increased GRP78 ubiquitylation in a concentration-dependent manner (Fig. 4D). To verify that the increased GRP78 ubiquitination by OTUDin3 was due to targeting OTUD3, we performed endogenous GRP78 ubiquitination assays in OTUD3 KO cells. The results showed that OTUDin3 had little effect on GRP78 ubiquitination in OTUD3 KO cells (Fig. 4E). Collectively, OTUDin3 increased GRP78 ubiquitylation by inhibiting the deubiquitinating activity of OTUD3.
A previous study showed that the N-terminal OTU domain of OTUD3 mediated the physical interaction with GRP78 [26]. Given that OTUDin3 binds to the OTU domain, we further investigated whether OTUDin3 disrupted the interaction between OTUD3 and GRP78 by performing co-IP and GST-pulldown assays. Co-IP and GST-pulldown assays con rmed that OTUDin3 did not affect the interaction between OTUD3 and GRP78 (Fig. 4F, Supplementary Fig. 4A). In addition, OTUD3 formed homodimers to perform deubiquitinating activity [23]. Co-IP and GST pull-down assays showed that OTUDin3 also did not affect the dimerization of OTUD3 (Fig. 4G, Supplementary Fig. 4B).
Taken together, OTUDin3 enhanced degradation of GRP78 by inhibiting the deubiquitinating activity of OTUD3 and increasing GRP78 ubiquitylation. These results are consistent with the expected mechanism of an OTUD3 inhibitor.

OTUDin3 inhibits NSCLC growth in vivo
To further investigate the effects of OTUDin3 on NSCLC in vivo, we next performed an H1299 xenograft mouse model experiment in which nude mice were implanted with H1299 cells. When tumors reached approximately 150-250 mm 3 , 18 tumor-bearing mice were divided into 6 blocks according to tumor size, and then the mice in each block were randomly assigned to different groups. The treatment groups were administered 10 mg/kg and 20mg/kg of OTUDin3 by intraperitoneal injection every other day, respectively. Under identical experimental conditions, the control group was administered corresponding volume of solvent. The results showed that treatment with OTUDin3 led to a signi cant inhibition of tumor growth (Fig. 5A). Compared with the control group, tumor volume and weight were signi cantly reduced in the treatment groups ( Fig. 5B-D). Immunohistochemical results showed that compared with the control group, the levels of GRP78 and Ki67 in the treatment groups decreased, and the level of Cleaved Caspase-3 increased (Fig. 5E). Meanwhile, there was no signi cant difference in average body weight between the control and treatment groups (Fig. 5F). Concomitantly, H&E staining showed that OTUDin3 had no signi cant effect on the major organs including heart, liver, spleen, lung and kidney ( Fig. 5G). Summarizing, OTUDin3 inhibited the NSCLC growth in vivo with no obvious health problems or weight loss of mice.

Discussion
As critical components in the regulation of protein homeostasis, DUBs are extensively involved in cell cycle regulation, DNA damage repair and cell growth control, which are all hallmarks of tumorigenesis [41]. Mechanistically, most cancers are characterized by overexpression of oncoproteins, which are normally degraded by the ubiquitin-proteasome pathway. However, when DUBs are aberrantly activated genetically or functionally, degradation of their corresponding oncoprotein substrates may be reduced, leading to oncoprotein accumulation [42]. Therefore, inhibition of the aberrantly activated DUBs could be a novel and promising anti-cancer strategy. Currently, an increasing number of potent inhibitors targeting oncogenic DUBs have been developed [37]. A variety of DUB inhibitors have been veri ed to be effective in malignant tumors, especially those targeting USP1, USP7, and UPS14, displaying great potential for clinical application, such as ML323, SJB3-019A, KSQ-4279, Compound 4, FT671 and VLX1570 [14,16,17,38,43,44]. Most encouragingly, KSQ-4279, a rst-in-class USP1 inhibitor, has entered a Phase clinical study for patients with advanced solid tumors (NCT05240898). In the meanwhile, the number of inhibitors targeting non-USP DUBs is limited. Although the research and development of DUB inhibitors has made great progress in recent years, it is still in its infancy, and the DUB inhibitor eld still has great potential for drug development.
The deubiquitylase OTUD3 was initially demonstrated to stabilize the tumor suppressor PTEN and play a suppressive role in breast tumorigenesis [23]. However, further study showed that Otud3 transgenic mice were more susceptible to Kras G12D -driven lung cancer while Otud3 KO mice were less susceptible, implicating an oncogenic role of OTUD3 in lung cancer [26]. Meanwhile, OTUD3 was found to be highly expressed in human NSCLC tissues, and the high expression of OTUD3 correlated with poor survival in NSCLC patients [26]. Mechanistic studies indicated that OTUD3 promoted lung tumorigenesis by deubiquitylating and stabilizing GRP78 [26]. GRP78 is a potent anti-apoptotic protein and plays a vital role in tumor cell survival, tumor progression and angiogenesis, metastasis, and resistance to therapy [45]. In addition, a study on CHIP, an E3 ubiquitin ligase for OTUD3, further con rmed that OTUD3 promoted NSCLC through an OTUD3-GRP78 signaling axis [46]. Thus, OTUD3 is a potential drug target for anticancer therapy in NSCLC. In addition, Otud3 KO in mice did not cause strong adverse effects, such as lethality or developmental defects [23]. Taken together, inhibition targeting OTUD3 is feasible and safe.
In this study, we identi ed lead compounds using virtual screening against a diverse set of 100 000 compounds. Virtual screening technology is the core of CADD. With the rapid increase in the availability of protein and small-molecule databases, the applicability of CADD has been extended to nearly every stage of drug discovery, including target identi cation and validation, lead discovery and optimization and preclinical tests [47]. Compared with traditional methods, CADD increases the hit rate of drug screening by more than 100 times, from ~ 0.01% to 1 ~ 2% [48]. Due to the high hit rates, CADD is becoming the fundamental basis of industrial drug discovery as well as academic research [49]. Currently, DUB inhibitors were identi ed mainly based on HTS, and the most frequently used substrates for DUB activity pro ling are ubiquitin-7-amino-4-methylcoumarin (Ub-AMC) and more recently Ub-rhodamine (Ub-Rho) [12]. Compared with HTS, virtual screening is more e cient and economical, which is an alternative strategy for inhibitors discovery of DUBs with known crystal structures.
After virtual screening and biological experiments, we identi ed OTUDin3 as an OTUD3 inhibitor. Then, we demonstrated that OTUDin3 inhibited NSCLC cell growth, migration and invasion, and induced apoptosis by inhibiting the deubiquitinating activity of OTUD3 and enhancing the ubiquitination and degradation of GRP78. Finally, an H1299 xenograft mouse model experiment also demonstrated that OTUDin3 inhibited lung tumor growth in vivo by downregulating the level of GRP78. Based on these ndings, we concluded that OTUDin3 inhibits NSCLC by targeting the OTUD3-GRP78 signaling axis (Fig. 6). Identi cation of OTUDin3 con rmed that targeting DUBs is a viable strategy for anti-cancer therapy. Our ndings are encouraging, and further studies aimed at resolving the co-crystal structure of OTUD3 complexed with OTUDin3 and optimization of OTUDin3 for further investigation are warranted and will be reported at a later stage.
OTUD3 plays context-dependent roles in tumorigenesis of different types of cancers [50]. In addition to regulating GRP78 to promote tumorigenesis in NSCLC, OTUD3 suppressed tumorigenesis in breast, colon, liver, brain, esophageal, and cervical cancer by stabilizing PTEN, ZFP36 or p53 [23][24][25]. Recently, a study showed that OTUD3 is an important regulator of glucose and lipid metabolism, and OTUD3 insu ciency is associated with obesity and a higher risk of diabetes [27]. In addition, OTUD3 prevented Parkinson's disease by stabilizing iron regulatory protein 2 [28]. Depletion of OTUD3 resulted in a disorder of iron metabolism, motor de cits, and nigrostriatal dopaminergic neurodegeneration, which resembled the pathology of PD[28]. Collectively, OTUD3 plays a positive role in these physiological processes, which means that an OTUD3 inhibitor may cause corresponding side effects. Therefore, it is necessary to consider drug delivery technology in the application of OTUD3 inhibitors in NSCLC treatment. At present, the research on targeted drug delivery for lung cancer has made progress [51]. The application of a targeted drug delivery system can be considered to avoid possible side effects of OTUDin3.
The good news is that OTUD3 plays a negative role in innate antiviral immunity via restricting innate antiviral immune signaling [52]. OTUD3 de ciency in mice results in enhanced innate immunity, a diminished viral load, and morbidity [52]. This implies that OTUDin3, may have potential application in antiviral therapy.
In summary, we identi ed the rst inhibitor of OTUD3 by virtual screening and biological experiments, which exhibited potent NSCLC inhibition both in vitro and in vivo. OTUDin3 enhanced degradation of the substrate GRP78 by inhibiting the deubiquitinating activity of OTUD3. Our ndings suggest that inhibition targeting OTUD3 may be a promising therapeutic strategy for NSCLC.

Computational virtual screening
The three-dimensional structures of OTUD3 (PDB ID: 4BOU) and OTUD5 (PDB ID: 3TMP) were obtained from the Protein Data Bank (http://www.rcsb.org/). The structure of OTUD3 was prepared before molecular docking using AutoDockTools-1.5.7. The GridBox was de ned to include S1 Ub binding site. A diversity set of 100 000 Screening Compounds in SDF format were obtained from the ChemDiv (http://www.chemdiv.org/). The SDF format les were converted to PDB format using Open Babel 3.1.1. The converted PDB les were then converted to PDBQT format using a ligand preparation script from AutoDockTools-1.5.7. Then, AutoDock Vina 1.1.2 was used for the subsequent molecular docking. Detailed procedures are described in the protocol [53]. The small molecules were ranked based on the a nity energy.
All of the visualization of the structure les was performed using PyMOL Molecular Graphics System, Version 2.6.0a Schrödinger, LLC. Protein-ligand interactions were analyzed using the Protein-Ligand Interaction Pro ler web tool.
Cell lines and culture conditions. The cell lines H1299, H1650, H460, A549, HPAEpiC and HEK293T were obtained from the American Type Culture Collection (ATCC), authenticated by STR pro ling (Promega) and shown to be mycoplasma-free using a mycoplasma test. H1299, H1650, H460 and HPAEpiC cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. A549 and HEK293T cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. All of the cells were grown in a humidi ed environment with 5% CO 2 at 37°C. Medium and supplements were purchased from Gibco except where indicated.
Reagents, recombinant proteins and antibodies.

Lentivirus infection
To generate the lentiviral shRNA constructs against human OTUD3, the OTUD3 shRNA sequences were cloned into the pCDH-puro vector. The shRNA sequences were reported in previous research [26]. The viruses were used to infect cells in the presence of polybrene. Forty-eight hours later, H1299 cells were cultured in medium containing puromycin for the selection of stable clones. The clones that stably knocked down OTUD3 were identi ed and veri ed by western blotting. The OTUD3 shRNA sequence was: 5′-TGGAAATCAGGGCTTAAAT-3′.

Cell viability assays
Cells were seeded in a 96-well plate at a density of 5000 cells per well. Twenty-four hours later, cells were treated with DMSO or the indicated compounds at varying concentrations for 72 h, and then assessed for viability using a CCK-8 Kit (CK001-500T, LABLEAD). For the CCK-8 assay, 10 µL of CCK-8 which was dissolved in 90 µL of medium was added to each well. The absorbance at 450 nm was measured by a microplate reader (PerkinElmer) after being incubated at 37°C for 1 h. IC 50 values were obtained using GraphPad Prism with nonlinear regression.

Colony formation assays
The cells were diluted to a single-cell suspension and 1000 cells were cultured in each well of a six-well plate at 37°C with 5% CO 2 incubator for 24 h. Then cells were treated with different concentrations of OTUDin3 for 10-14 days until clones were clearly visible. Cell culture plates were gently washed by PBS twice, xed with a 4% paraformaldehyde phosphate buffer for 15 min and then stained with crystal violet for 15 min. Washing the excess stain with ddH 2 O and drying them at room temperature for several hours.
Photographs of the stained colonies were taken.

Wound-healing assays
Wound-healing assays were performed to evaluate the mobility of the cells. Cells were seeded in 6-well plates. After 24 h of culture, each well was manually scratched with a 200 µL pipette tip, and washed with PBS three times. Then, the cells were cultured in basic RPMI-1640 medium without FBS, and treated with varying concentrations of OTUDin3. The scratch area was photographed 24 h later. The distance between two cell edges was analyzed using ImageJ software (NIH).

Invasion assays
A cell invasion assays were performed in a 24-well Transwell plate with 8.0 µm polycarbonate membrane inserts (Corning) coated by 60 µL Matrigel matrix (Corning). In brief, 2 × 10 5 H1299 cells resuspended in serum-free medium with varying concentration of OTUDin3 were plated in each insert. The inserts were placed in the well containing RPMI 1640 medium with 10% FBS. After 24 h, non-invaded cells were removed and the inserts were washed in PBS, xed in 4% formaldehyde for 20 min and stained with 0.1% crystal violet for 20 min. The wells were photographed and the stained cells were counted.

Flow cytometry-based apoptosis detection
Flow cytometry-based apoptosis assays were performed using an Annexin V-FITC/PI Apoptosis Detection kit (P04D02, Gene-Protein Link). H1299 cells in 6-well plates were treated with different concentrations of OTUDin3 for 24 h. The cells were collected by centrifugation and resuspended in 100µL of binding buffer. Then, 5 µL of Annexin V-FITC and 5 µL of propidium iodide (PI) were added and incubated at room temperature away from light for 10 min. Then, 400 µL of binding buffer was added, and samples were detected by ow cytometry within 1 h. The data were processed by FlowJo_V3.8.1 software.

OTUD3 KO cell line construction
In a 6-well tissue culture plate seed 1.5×10 5 − 2.5×10 5 cells in 3 ml of standard growth medium per well, 24 h prior to transfection. Grow cells to a 40-60% con uency. Initial cell seeding and cell con uency after Co-immunoprecipitation assays HEK293T cells were transfected with indicated plasmids using Polyethylenimine (PEI, Invitrogen) reagent according to the manufacturer's protocol, and treated with DMSO or different concentrations of OTUDin3.
For immunoprecipitation assays, cells were lysed with HEPES lysis buffer (20 mM HEPES, pH 7.2, 50 mM NaCl, 0.5% Triton X-100, 1 mM NaF and 1 mM dithiothreitol) supplemented with a protease-inhibitor cocktail (Roche). Immunoprecipitations were performed using the indicated primary antibody and protein A/G agarose beads (Santa Cruz) at 4°C. The immunocomplexes were then washed with HEPES lysis buffer four times. Both lysates and immunoprecipitates were examined using the indicated primary antibodies followed by detection with the related secondary antibody and the SuperSignal West Pico chemiluminescence substrate (Thermo).
Bacterially expressed GST and GST-OTUD3 bound to glutathione-Sepharose 4B beads (from GE) were incubated with Myc-GRP78 or Flag-OTUD3 expressed in HEK293T cells for 4 h at 4°C in the presence of DMSO or different concentrations of OTUDin3. Then the beads were washed with GST binding buffer (100 mM NaCl, 10 mM Tris, 50 mM NaF, 2 mM EDTA, 0.5 mM Na3VO4 and 1% NP40) four times and proteins were eluted, followed by western blotting or Coomassie blue staining.

NSCLC xenograft study
The animal experimental protocols were approved by the Animal Care and Use Committee of Beijing Institute of Lifeomics, China. Male BALB/c nude mice (6-to-8 week old) purchased from Vital River Laboratory Animal Technology Co, were subcutaneously inoculated with 6 × 10 6 H1299 tumor cells in 1:1 mixture of RPMI-1640 and Matrigel (Corning). The e cacy study was initiated when tumors reached approximately 150-250 mm 3 . Using randomized block design, 18 tumor-bearing mice were divided into 6 blocks according to tumor size, and then mice in each block were randomly assigned to different groups.
Mice received a dosing regimen of 10 or 20 mg/kg OTUDin3 every other day (intraperitoneal injection), and 10% DMSO, 10% polyoxyethylene castor oil and 80% saline served as a vehicle control. Body weight and tumor volume of the mice were monitored every other day. The length and width of tumors were measured with a vernier caliper. Tumor volume was calculated by the formula: V = 0.5 × length × width 2 .
When the maximum tumor volume of mice reached 2000 mm 3 , tumor samples were collected. H&E staining and immunohistochemistry were performed by Wuhan Servicebio Technology Co.,Ltd. All animals were handled in strict accordance with the "Guide for the Care and Use of Laboratory Animals" and the "Principles for the Utilization and Care of Vertebrate Animals", and all of the animal work was approved by the Institutional Animal Care and Use Committee at the Beijing Institute of Lifeomics. Studies were tailored to minimize the number of animals used, yet allow su cient numbers to address any variability in drug exposure. For e cacy studies where growth rate differences introduce some variability, n = 6 engrafted animals per treatment cohort provided su cient statistical power for a wellbehaved xenograft model. Owing to the need to monitor potential adverse effects with rst in-life assessment of novel compounds, no study was conducted under blinded conditions. Statistical analysis.
Statistical analyses were performed using GraphPad Prism 7.0 software. All results are shown as mean ± standard deviation (SD) of multiple independent experiments. Normality and lognormality test was used.
If data did not meet the criteria for normal distribution, nonparametric test was performed. Unpaired Student's t-tests were used for two-group comparisons. Two-way ANOVA followed by Fisher's LSD test was applied to estimate how the mean of a quantitative variable changes according to the levels of two categorical variables. Wilcoxon test were used for the paired samples that are not normally distributed in mice experiment. P values < 0.05 were considered statistically signi cant.

Con ict of Interest
The authors declare no con ict of interest.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.  OTUDin3 inhibits NSCLC cell growth, migration and invasion, and induces apoptosis by targeting OTUD3.
A. OTUD3was knocked down by shRNA in H1299 cells. The protein levels of OTUD3 were analyzed by western blotting.
B. Cell proliferation assays in H1299 shNC and H1299 shOTUD3 cells treated with increasing concentrations of OTUDin3 for 72 h. Viability of cells was determined by the CCK-8 kit.  The apoptotic cell population including early and late apoptotic cells was quanti ed using an apoptotic kit and analyzed by ow cytometry. The data were analyzed by FlowJo_V10.8.1 software (Treestar, shland, OR).
All the data shown are mean ± SD. n = 3 or 5 independent experiments. Two-sided P values were calculated using two-way ANOVA followed by Fisher's least signi cant difference test. *p < 0.05, **p < 0.01, ***p< 0.001, ns = no signi cance, H1299 shNC vs H1299 shOTUD3. OTUDin3 downregulates GRP78 by inhibiting the deubiquitinating activity of OTUD3 A. H1299 cells treated with 5 μM OTUDin3 for the indicated time were lysed for western blotting analysis probing for GRP78, OTUD3, and GAPDH as indicated. OTUDin3 inhibits the NSCLC growth in vivo A. Photographic images of xenograft tumors from control and OTUDin3 treatment groups. Tumor-bearing mice were divided into 6 blocks according to tumor size, and then mice in each block were randomly assigned to different groups (Randomized block design).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.