GRK3 is a poor prognosticator and serves as a therapeutic target in advanced gastric adenocarcinoma

Background G protein-coupled receptor (GPCR) is the most targeted protein family by the FDA-approved drugs. GPCR-kinase 3 (GRK3) is critical for GPCR signaling. Our genomic analysis showed that GRK3 expression correlated with poor prognosis of gastric adenocarcinoma (GAC) patients. However, GRK3’s functions and clinical utility in GAC progression and metastases are unknown. Methods We studied GRK3 expression in normal, primary, and metastatic GAC tissues. We identified a novel GRK3 inhibitor, LD2, through a chemical-library screen. Through genetic and pharmacologic modulations of GRK3, a series of functional and molecular studies were performed in vitro and in vivo. Impact of GRK3 on YAP1 and its targets was determined. Results GRK3 was overexpressed in GAC tissues compared to normal and was even higher in peritoneal metastases. Overexpression (OE) of GRK3 was significantly associated with shorter survival. Upregulation of GRK3 in GAC cells increased cell invasion, colony formation, and proportion of ALDH1+ cells, while its downregulation reduced these attributes. Further, LD2 potently and specifically inhibited GRK3, but not GRK2, a very similar kinase to GRK3. LD2 highly suppressed GAC cells’ malignant phenotypes in vitro. Mechanistically, GRK3 upregulated YAP1 in GAC tissues and its transcriptional downstream targets: SOX9, Birc5, Cyr61 and CTGF. Knockdown (KD) YAP1 rescued the phenotypes of GRK3 OE in GAC cells. GRK3 OE significantly increased tumor growth but LD2 inhibited tumor growth in the PDX model and dramatically suppressed peritoneal metastases induced by GRK3 OE. Conclusions GRK3, a poor prognosticator for survival, conferred aggressive phenotype. Genetic silencing of GRK3 or its inhibitor LD2 blunted GRK3-conferred malignant attributes, suggesting GRK3 as a novel therapeutic target in advanced GAC. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-022-02463-6.


Background
Gastric adenocarcinoma (GAC) is the fifth most common cancer and the third leading cause of cancer-related mortality worldwide [1]. GAC is usually diagnosed in the late stages, and metastatic GAC is invariably incurable [2][3][4]. Despite some advances in therapies such as surgery, systemic therapy, and radiation therapy, GAC mortality rates remain high. Advanced genetic and molecular profiling has yielded a vast quantity of new information for potential therapeutic exploitation and risk stratification [5,6]. However, novel predictive and prognostic biomarkers to serve as therapeutic targets are still limited.
Through shRNA and cDNA functional screening of human kinases, we previously identified that GRK3 (G protein-coupled receptor kinase 3) is an essential kinase for prostate cancer progression [7]. Yet, the mechanism of its actions in cancer progression is still largely unclear. GRK3 is best known to phosphorylate the agonist-occupied form of the β-adrenergic and related G protein-coupled receptors, leading to broad regulation of receptor functions. G protein-coupled receptors (GPCRs) play a central role in signal transmission, thereby controlling many facets of cellular functions. Increasing evidence suggests a role of GPCRs and their ligands in different aspects of tumor biology [8].
GRK3 shows varied expression levels and functions in different tumor types. For example, downregulation of GRK3 correlated with increased growth of glioblastoma cells [9] and was associated with worse outcomes in pancreatic ductal adenocarcinoma [10] and hepatocellular carcinoma [11]. In contrast, over-expression of GRK3 in prostate tumor cells and colon cancer promoted tumor growth and metastases through the induction of angiogenesis [7,12]. Therefore, the roles of GRK3 in various cancers are likely context dependent. However, the role of GRK3 in GAC remains unclear and needs to be pursued.
In this study, we evaluated the expression of GRK3 in primary GAC tissues and paired adjacent normal tissues as well as in peritoneal carcinomatoses (PC) cells from GAC patients. We then elucidated the functional role of GRK3 in promoting tumor growth and metastases using genetic approaches. Mechanistically, we identified crosstalk between GRK3 and YAP1, a key player in GAC progression and metastases [13]. Importantly, we identified a novel GRK3 inhibitor, LD2, through a chemical library screen and demonstrated its strong anti-tumor activity in vitro and in vivo in patient-derived PC cells with high GRK3 and YAP1 expressions. Thus, GRK3 could serve as a prognostic biomarker and novel therapeutic target in GAC.

Cells and reagents
The human GAC cell lines AGS, MKN45, SK-GT-5 (GT-5), KATO III, SNU-1, and SNU-16 were purchased from the American Type Culture Collection (ATCC, Manassas, Virginia) and previously described [14]. The immortalized normal gastric epithelial cell lines GES-1 and HFE145 were described previously [15,16]. GA0518 patient-derived cells were isolated from a patient-derived xenograft (PDX) model implanted with PC cells from a GAC patient [14]. Patient-derived GA0804, GA0825, GA0515, and GA0313 cells were enriched directly from patient's PC samples after lysing red blood cells. These cells were grown in RPMI 1640 medium with 10% fetal bovine serum at 37 °C in 5% CO 2 . All human cell lines were authenticated in the Cytogenetics and Cell Authentication Core facility of The University of Texas MD Anderson Cancer Center every 6 months and tested for mycoplasma in our laboratory every 6 months. Antibodies used included anti-GRK3 rabbit antibody (Abcam, Cambridge, United Kingdom), anti-YAP1 rabbit antibody (Cell Signaling Technology, Danvers, MA), anti-SOX9 rabbit antibody (EMD Millipore, Billerica, MA), anti-CTGF mouse antibody (Santa Cruz Biotechnology, Santa Cruz, CA), anti-Cyr61 rabbit antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and anti-Survivin rabbit antibody (Cell Signaling Technology, Danvers, MA). LD2 was identified by screening chemical libraries for inhibition of GRK3's kinase activity, performed by Dr. Kevin Dalby's laboratory at The University of Texas at Austin and Dr. Wenliang Li at The University of Texas Health Science Center at Houston. LD2 was provided by Dr. Wenliang Li for the experiments in this study.

Patient samples and tumor tissue microarrays (TMAs)
Ethical approval for this study was obtained from the ethics committee of China Medical University (CMU) and Conclusions: GRK3, a poor prognosticator for survival, conferred aggressive phenotype. Genetic silencing of GRK3 or its inhibitor LD2 blunted GRK3-conferred malignant attributes, suggesting GRK3 as a novel therapeutic target in advanced GAC. Keywords: Gastric adenocarcinoma, GRK3, Hippo/YAP1, Prognosis, Peritoneal metastases the Institutional Review Board of MD Anderson Cancer Center (MDACC). All patients who volunteered to provide research specimens signed an approved written consent document. Experiments were carried out at CMU and MDACC on respective tissue resources available at each institution.
Twenty pairs of GAC and normal tissues were from the First Hospital of CMU, and 3 pairs of tissues as well as PC samples from 42 cases were from the Department of Gastrointestinal Medical Oncology at MDACC for quantitative real-time polymerase chain reaction (qPCR). The 42 PC samples for immunofluorescent staining were from the Department of Gastrointestinal Medical Oncology at MDACC. The TMA included a total of 422 patients who underwent total or subtotal gastrectomy with lymphadenectomy for GAC from January 2009 through December 2014 from the Department of Surgical Oncology and General Surgery at First Hospital of CMU. Among 422 patients, 393 patients have complete clinical survival outcomes that were analyzed for survival association for GRK3 expression. Written informed consents were obtained from all patients. None of the patients had received chemotherapy or radiotherapy before their surgical procedure. Detailed postoperative pathology reports and other demographic and clinical data were obtained, including age, gender, tumor location, tumor size, differentiation status, growth pattern, invasion depth, lymph node metastasis, distant metastases, TNM stage, and vein invasion. We used the TNM classification for GAC based on the 7th edition of the AJCC staging manual. All patients were followed up via telephone inquiries or questionnaires. The follow-up time was 2-98 months (median 51 months).

Immunohistochemistry
For formalin-fixed, paraffin-embedded TMA, 5-μm-thick tissue sections were deparaffinized in xylene, followed by dehydration in an ethanol series. The slides were subjected to antigen retrieval with 10 mM sodium citrate (pH 6.0) for 45 min. Before staining, non-specific binding was blocked by incubation with hydrogen peroxide as a peroxidase suppressor (Thermo Fisher Scientific, Waltham, MA) and normal horse serum (Vector Laboratories, Burlingame, CA) as a blocking buffer, followed by incubation with 1:100 anti-GRK3 (ab109303, Abcam) antibodies in antibody diluent (BioGenex, San Ramon, CA) at 4 °C overnight. All sections were briefly washed with PBS and incubated at room temperature with horseradish peroxidase-conjugated secondary anti-rabbit antibody. The color was then developed by incubation with a DAB substrate kit (Vector Laboratories). Nuclei were counterstained blue with hematoxylin (Sigma-Aldrich, St. Louis, MO) and mounted in VectaMount Permanent mounting medium (Vector Laboratories). Isotype immunoglobulin G controls were used as negative controls for the staining.
Two pathologists who were blinded to patient outcomes independently interpreted the immunostaining results using a semi-quantitative scoring system. Immunostaining reactions were evaluated by staining intensity (0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining) and the percentage of stained cells (0, ≤ 5%; 1, 5-25%; 2, 25-50%; 3, 50-75%; 4, > 75%). Then, the percentage of positive cells and the staining intensity were multiplied to generate the immunoreactivity score (IS) for each case. If there were discrepancies in the IS as determined by the two pathologists, specimens were rescored until a consensus was reached. Then the cutoff value was defined by the ROC curve. The cutoff value of IS was 1, so the IS = 0 was defined as negative and IS > = 1 as positive.

Generation of GAC cells with genetic knockdown or overexpression of GRK3
GRK3 gene knockdown (KD) by the lentiviral CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) system was performed following instructions on the Massachusetts Institute of Technology website (http:// crispr. mit. edu/, which is discontinued now). A second software for designing guide RNAs (gRNAs) was obtained from the German Cancer Research Center (http:// www.e-crisp. org/E-CRISP/ desig ncris pr. html) gRNA design guidelines. The designed pairs of gRNA (Suppl Table 2) formed duplexes which were then ligated using T4 ligase to V2mO (pLenticrispr-V2-mOrange) (#140,206, Addgene, Cambridge, MA) precut by BsmB1 and gel purified. The ligates were transformed into Stbl3 competent cells. Clones on LB plates were screened, and the positive ones were verified by sequencing. The plasmids with desired gRNAs were then packaged into lentiviruses using psPax2 and pDM2.G at a ratio of 10:10:1 in HEK293T cells. Lentiviruses were transduced into desired target cell lines by infecting with polybrene at 8 μg/ml. Stable cell lines were obtained upon selection with puromycin at concentrations determined by its kill curves in different cell lines. The efficacy of gene KD in target cell lines was determined by Western blots and qPCR for further experiments.

qRT-PCR
Total RNA was extracted from GAC cells treated with LD2 at indicated dosage using Direct-zol RNA Kit (Zymo Research), processed for cDNA synthesis using the Reverse Transcription Kit (Applied Biosystems), and subjected to the qRT-PCR using SYBR Green Master Mix (Applied Biosystems). The expression level of indicated genes was normalized to the expression of GAPDH as housekeeping gene. Primer sequence were listed in Suppl Table 3.

Matrigel invasion assay
The invasion assay was performed as described previously [19]. Briefly, 1 × 10 5 cells of MKN45 or KATO III with overexpression of GRK3 compared with control or GA0518, GA0814 and AGS cells were plated on the well of transwell insert (Corning) and then 600 µl of culture medium with 20% FBS was gently added to the lower well followed by incubation of the cells with or without treated with LD2 at indicated dosage in 60 µl of culture medium containing 0.5% FBS at 37 °C, 5% CO2 for 48 h. The culture media and cells inside of the insert were removed with cotton-tipped applicators and the migrated cells were fixed with 4% formaldehyde and stained with 0.2% crystal violet. The migrated cells were counted under an invert microscope.
Tumor sphere formation assay GA0518, GA0804 and AGS cells (800/well) were seeded in triplicate onto a 24-well ultra-low attachment plate (Corning, Corning, NY) in serum-free DMEM /F-12 supplemented with 10 ng/ml epidermal growth factor, 5 mg/ml insulin, 0.5 mg/ml hydrocortisone, and bovine pituitary extract (Invitrogen). The cells were then treated with control DMSO or LD2 at the indicated doses. After 10-14 days of culture, tumor spheres formed (diameter > 100 μm) were counted under a microscope.
Flow cytometry ALDH1 + labeling ALDH1 activity was assessed by flow cytometry using an ALDEFLUOR kit (STEMCELL Technologies, Vancouver, Canada) per the manufacturer's instructions. Cultured human GAC cells and PC cells were suspended in ALDEFLUOR assay buffer at 1 × 10 6 cells/ml and added to a tube containing 5 μl of the ALDH1 substrate. As a negative control, a 0.5-ml aliquot from each sample was treated with ALDH1-specific inhibitor. After 30 min of incubation at 37 °C and then centrifugation, the cells were washed with the assay buffer, followed by resuspension in 0.3 ml of ice-cold ALDEFLUOR assay buffer for flow labeling (FACS Calibur, BD Biosciences).

In vivo xenografts and PDX
In vivo experiments were conducted in accordance with the Institutional Animal Care and Use Committee. MKN45 GAC cells with GRK3 overexpression (GRK3 OE) and GFP control cells (NC) were subcutaneously injected into severe combined immunodeficient (SCID) mice (n = 5/group) and then monitored tumor growth for three weeks. After three weeks, tumor burden, tumor weight and tumor volume were measured as previously [13]. To determine the antitumor activity of LD2 in vivo, we applied a PDX model using GA0518 patient-derived cells., 5 × 10 6 GA0518 cells were subcutaneously injected into severe combined immunodeficient (SCID) mice (n = 5/group, two injections per mouse on both flanks). After 7-10 days, the mice were injected with vehicle control DMSO: PBS (1:1) (100 μl/mouse) or LD2 (20 mg/kg) 5 times per week for 2 weeks. Tumor size was measured with a digital caliper (VWR International, Radnor, PA) once tumors reached a visible size, and tumor volume in mm 3 was determined by the formula [(length × width) 2 ] × 0.52, with length and width in mm. Mouse tumor weight and body weight were measured as described previously [20]. For the PC metastasis model, SCID mice were peritoneally injected with 1 × 10 6 MKN45 GFP, MKN45 GRK3 OE, MKN45 GRK3 OE treated with LD2. 1 week after tumor cell injection, mice were randomly grouped. LD2 (20 mg/kg) was treated 5 times per week for 2 weeks. The tumor luminescent density was monitored every week. All these measurements were compared using the unpaired Student t-test.

Statistical analysis
Student t-test and Fisher exact test were used to analyze colony formation and cell migration assays. Kaplan-Meier method was used to estimate the probability of survival. Log-rank test and Cox regression analysis were used to determine the association between markers and survival outcomes using SPSS 22.0 (IBM, Armonk, NY). Other assays are presented in graphs as mean ± standard error of the mean and represent the results of at least 3 experiments. The significance of differences between groups was judged using the 2-tailed Student t-test. Results were considered statistically significant if the P-value was less than 0.05. All statistical tests were performed using GraphPad Prism 8 software (GraphPad Software, Inc., San Diego, CA).

Expression of GRK3 was increased in primary and metastatic GACs and associated with aggressive phenotype, and shorter survival
To examine the expression status of GRK3 in GAC, we first explored TCGA dataset and found that expression of GRK3 was higher in gastric tumor tissues than that in normal tissues (Fig. 1A). Expression of GRK3 in tumor tissues in a large GAC cohort (n = 876) significantly associated with poor survival as well as a higher risk of death (P = 1.2e-08; hazard ratio = 1.74, confidence interval 1.43-2.11, kmplot.com) (Fig. 1B). To confirm these findings, we performed qPCR for GRK3 in 23 pairs of GAC tissues and adjacent normal tissues. Primers of qPCR for detecting GRK3 were listed in Supplementary Table 2. In 15 of 23 pairs (65%), GRK3 expression was higher in GAC tissues than in the adjacent normal tissues (Fig. 1C). To further validate this observation, we performed immunohistochemical (IHC) staining on a GAC TMA containing 393 pairs of GAC and normal tissues. GRK3 was typically expressed in the cytosol and cell membrane (Fig. 1D). Compared with the paired normal tissues, GRK3 expression was higher in GAC tissues. Both GAC phenotypes (intestinal and diffuse) and signet ring cell type of diffused GAC were positive for GRK3 (red arrow) (Fig. 1D). The positive expression rate for GRK3 was 61.1% (240/393) in GACs, but only 32.3% (127/393) in normal tissues. Intriguingly, GRK3 expression in the primary GACs was statistically associated with the higher rate of lymph node metastases, diffused type of GAC, and a higher TNM categories (Supplementary Table 3). The Kaplan-Meier analysis showed that GAC patients with higher GRK3 expression had shorter overall survival (P = 0.002, Fig. 1E). Univariate analysis found that GRK3 positive expression was significantly associated with high risk of death with HR = 1.563 and 95% confidence interval (CI) between 1.176-2.079 and P value was 0.002, although the multivariate analyses found that GRK3 is not an independent prognostic factor for survival (Supplementary Table 4). On further analysis, overexpression of GRK3 was associated with a higher AJCC stage and aggressive phenotype ( Fig. 1F and G).

GRK3 is highly expressed in peritoneal carcinoma (PC) cells from metastatic GAC patients
Based on the results in TMAs from our GAC cohorts, where GRK3 over-expression correlated with aggressive phenotype and lymph node metastases (Fig. 1G), we sought to determine the expression patterns of Fig. 1 The expression of GRK3 is upregulated in primary and metastatic GACs, which is associated with aggressive phenotypes and shorter survival. A Expression of GRK3 is shown in GAC tumor tissues compared with normal tissues from TCGA STAD and GTEX cohort (****P < 0.0001). B Expression of GRK3 is associated with poor survival of 876 GAC patients in KM plotter database (P = 1.2e-08). C Expression of GRK3 mRNA was measured by quantitative polymerase chain reaction (qPCR) and normalized to GAPDH in 23 pairs of primary GAC tissues and adjacent normal tissues. (**P < 0.001). D Immunohistochemistry (IHC) staining was performed using the GRK3 antibody in GAC tissue microarrays (TMAs) of primary tumor tissues and adjacent normal tissues. E Kaplan-Meier analysis of overall survival time according to GRK3 expression level in GAC TMAs. GAC patients with high GRK3 expression showed worse survival than patients with low expression did (P = 0.002). F&G The expression of GRK3 was analyzed with stage and lymph node metastases in our GAC cohort in TMAs. H GRK3 mRNA level was detected on normal tissues, primary tissues, and ascites cells by qPCR. **P < 0.001. I. GRK3 expression was detected using immunofluorescent staining in 22  GRK3 in PC samples of GAC. First, we subcutaneously injected several patients' PC cells (malignant ascites cells) into nude mice to generate PDXs as shown in Fig. S1A. Some PC cells (GA0518) formed tumors at the primary injection site (the primary tumor), and some mice further developed PC with ascites. We collected paired primary tumor cells and PC cells from the same mouse and determined GRK3 expression using Western blotting. GRK3 expression was higher in 3 PC samples compared with the paired primary tumors ( Supplementary Fig. 1A, right panel). To confirm that the expression of GRK3 is higher in PC cells, we performed qPCR on primary tumors, normal adjacent tissues, and PC samples from 28 GAC patients. We found that GRK3 mRNA levels were upregulated in the primary tumors (P = 0.0042) and PC cells (P < 0.001, Fig. 1H Fig. 1I and Supplementary Fig. 1B. Furthermore, in 4 pairs of the primary tumors and PC cells, each pair from the same patient, GRK3 expression was higher in 3 PC samples (75%) than in the corresponding primary tumor (Supplementary Fig. 1C). Altogether, these results suggest that the expression of GRK3 is increased in PC cells and may play an important role in GAC progression and metastases.

Upregulation of GRK3 in GAC cells augmented GAC cell growth and invasion, while downregulation of GRK3 inhibited malignant behavior of tumor cells in vitro
Next, we investigated the functional role of GRK3 in GACs using genetic modulations. First, we assessed GRK3 expression in the indicated cell lines using western blotting ( Fig. 2A). We found that protein levels of GRK3 in 6 GAC cell lines and 4 patient-derived PC cells (from malignant ascites) were much higher than in 2 normal gastric epithelial cell lines ( Fig. 2A). Then, we overexpressed GRK3 by transfected GRK3 cDNA into MKN45 cells with low GRK3 and confirmed the successful transfection of GRK3 in MKN45 cells using western blot and q-PCR (Fig. 2B). We found tumor cell invasion and colony formation were significantly increased in GRK3 overexpressed MKN45 cells compared to MKN45-GFP control cells ( Fig. 2C and D). Further, overexpression of GRK3 in MKN45 cells significantly increased tumor cell growth at any FBS concentration used (0-5%) media in 6 day (Fig. 2E) and (Supplementary Fig. 2A) and increased tumor sphere formation ( Supplementary Fig. 2B). Similarly, overexpression of GRK3 wild type cDNA in KATO III cells significantly increased its tumor sphere formation, invasion, colony formation and proliferation, which was reduced in its kinase dead mutant form (Fig. 2F, G  individual KD clones ( Supplementary Fig. 2G). Altogether, these data suggest that GRK3 plays an oncogenic role in GAC.

Identification of a GRK3 inhibitor LD2 that strongly suppressed the malignant behavior of GAC cells in vitro
Since GRK3 belongs to the kinase family, which is known to be druggable [21,22], we elected to investigate the potential of targeting GRK3 with inhibitors to block GAC progression. To this end, we screened a library of compounds for inhibiting GRK3 kinase using recombinant GRK3 protein and the Z-LYTE in vitro kinase assay from Invitrogen. Flow charts of our compound library screening strategies for identifying novel GRK3 inhibitors is shown in Supplementary Fig. 3. The primary screen was carried out using the Invitrogen Z-LYTE assay with single dose of 10 µM of compounds in the library. After the primary screen, hits were cherry picked and assayed for a dose response curve with 9 compound dilutions (twofold serial dilution) (separate manuscript in preparation). From these screens, we identified a compound, which we called LD2, its chemical structure was shown in Fig. 3A. We found that LD2 strongly inhibited GRK3's kinase activity with the half-maximal inhibitory concentration (IC50) of 82 nM (Fig. 2B). LD2's inhibition of GRK3 with IC50 of 82 nM was much stronger than inhibition of GRK2 with IC50 of 1954 nM (Fig. 2C), indicating the specificity of LD2 on GRK3. The specificity of LD2 targeting GRK3 was further confirmed when treating MKN45-GRK3 OE cells and control cells using LD2 at different dosage and found that LD2 significantly suppressed tumor cell growth in MKN45-GRK3 OE cells, but no effects in MKN45-GFP control cells ( Supplementary  Fig. 4A). Interestingly, we found that LD2 significantly suppressed tumor cell growth, colony formation,, and invasion,in a dose-dependent manner in patientsderived GA0518 and GA0804 cells, as well as AGS cell line (Fig. 3D-F,

and Supplementary Figs. 4B & 4C).
Most importantly, LD2 also suppressed GAC cells invasion induced by overexpression of GRK3 in both KATO III and MKN45 cells (Fig. 3G & 3H), suggesting that GRK3 played a critical role in regulating phenotypic aggressiveness of GAC, while LD2 could curtail the malignant phenotype induced by GRK3.

GRK3 was positively associated with YAP1 in GAC tissues
As demonstrated by us and others, the Hippo/YAP1 pathway plays an essential role in GAC progression, cancer stem cell (CSC) attributes, and PC formation [17,18]. To explore the relationship between GRK3 and YAP1 in GAC tumors, first, we found that there was a positive correlation between GRK3 and YAP1 mRNA level in TCGA STAD patients (Fig. 4A). Then, we measured the expression of GRK3 and YAP1 in 66 cases of GAC tissues and PC specimens using qPCR. Pearson correlation analysis showed that the expression of GRK3 was significantly associated with YAP1 expression (r = 0.3900, P = 0.0012; Fig. 4B). This correlation between GRK3 and YAP1 was further validated in GAC TMAs using IHC for GRK3 and YAP1 antibodies (Fig. 4C). Chi-square test analysis further indicated a strong positive association between GRK3 and YAP1 expression ( Fig. 4D and Supplementary Table 5). Further analyses of GRK3 staining on GAC TMA revealed that GRK3 was significantly associated with diffuse type of GAC which is more aggressive than intestinal subtype (Fig. 4E). Moreover, we have dualstained YAP1 and GRK3 in 49 PC metastatic cases and found that both GRK3 and YAP1 were positively correlated ( Fig. 4F; P < 0.0001). Further, overexpression of GRK3 in KATO III cells increased YAP1 transcription and protein expression ( Fig. 4G and J), while KD of GRK3 in GA0518 cells dramatically decreased YAP1 expression and transcription in 3 individual GRK3-KD clones by Western blotting and qPCR respectively, and these decreases corresponded to the decreased level of GRK3 ( Fig. 4H   and tumor sphere formation induced by GRK3 OR in both KATO III and MKN45 cell lines ( Supplementary  Figs. 5A-5D). Moreover, pharmacologic suppression of GRK3 using LD2 dramatically inhibited YAP1 protein expression and transcription in GA0518 cells (Fig. 4K). IF staining of YAP1 and GRK3 in GA0518 cells showed that expression of both YAP1 and GRK3 was dramatically decreased in cells treated with LD2 compared to that control cells (Supplementary Fig. 6A). In addition, LD2 significantly decreased YAP1 promoter-luciferase activity when GA0518 cells were transfected with YAP1 promoter-luciferase construct and then treated with LD2, which is in line with the reduction of YAP1 mRNA by LD2 (Supplementary Fig. 6B). Of note, LD2 also significantly inhibited YAP1/Tead transcriptional activity when GA0518 cells were co-transfected with Tead binding elements-luciferase construct and YAP1 cDNA, followed by LD2 treatment ( Supplementary  Fig. 6C). These results indicated that LD2 suppressed YAP1's own transcription and its downstream transcriptional activity.

GRK3 regulated YAP1 targets and inhibition of GRK3 reduced CSC attributes and aggressive phenotype of GAC cells
Previously, we and others have reported that Hippo/ YAP1 mediates CSC attributes and tumor progression through regulation of its downstream targets, such as SOX9, Birc5, and Cyr61 [17,23]. We postulated that GRK3 alters CSC attributes and GAC progression through YAP1-mediated downstream signaling. We first determined whether YAP1 down-stream targets, including SOX9, Birc5, Cyr61 and CTGF, were affected by GRK3 genetic alteration. As shown in Fig 5A and B, YAP1 targets, SOX9, Birc5, Cyr61 and CTGF, were significantly upregulated upon overexpression of GRK3 in both KATO III and MKN45 cells ( Fig. 5A and B, Supplementary Figs. 7A,7B and 7C). We further showed that SOX9 protein level was dramatically increased in both KATO III and MKN45 with overexpression of GRK3 (Fig. 5C), suggesting that GRK3 positively regulated SOX9, an important stem cell factor [24,25]. Intriguingly, pharmacological inhibition of GRK3 by LD2 significantly decreased YAP1 and its targets SOX9 and Cyr61 in a dose dependent manner in both GA0518 and GA0804 patient-derived PC cells (Fig. 5D and E).
To explore whether GRK3 inhibition affected the function of CSCs, tumor sphere formation was examined in 3 GAC cell lines with high GRK3 and YAP1 expression after treatment with LD2 at various concentrations. Tumor sphere formation in GA0518, GA0804 and AGS was significantly reduced by LD2 treatment in a dose-dependent manner (Supplementary Fig. 8A).
Of note, GRK3 overexpression in MKN45 and KATO III significantly increased tumor sphere formation in both GAC cells, which were dramatically suppressed by LD2 treatment in a dose dependent manner (Fig. 5F). ALDH1, a reliable CSC marker in tumor cells, reflects the CSC-like characteristics of various cancer types [26]. To examine the direct association of GRK3 with CSCs traits, we performed ALDH1 labeling in KATO III cells with or without GRK3 OE and found that ALDH1 + population is significantly increased in GRK3 OE cells compared to GFP-control cells while inhibition of GRK3 by LD2 dramatically decreased ALDH1 + population in three cell lines (Fig. 5G, 5H and Supplementary Fig. 8B).

GRK3 promoted tumor growth, while inhibition of GRK3 by LD2 significantly suppressed growth in the PDX model
To investigate the impact of GRK3 in GAC in vivo, we first determine the effects of overexpression of GRK3 in GAC cells in tumorigenicity. SCID mice were subcutaneously injected with MKN45 cells overexpressing GRK3 or GFP control (NC), which have been validated using qPCR and western blot (Fig. 2B).
Over the three-week of inoculation period, tumor growth was faster in GRK3 OE group compared to NC group. Tumor burden, tumor volume, and tumor weight at end of experiments were also significantly larger in GRK3 OE group than NC group (Fig. 6A-D). Correspondingly, Grk3 mRNA level was significantly increased in GRK3 OE tumors compared to NC tumors (Fig. 6E). Co-immunofluorescent staining (IF) further validated that expression of GRK3 and SOX9 was dramatically increased in tumor tissues from the GRK3 OE group, compared to the NC group (Fig. 6F).
To evaluate the anti-tumor activity of LD2 in vivo, we used GA0518 patient-derived PC cells which have high GRK3 and YAP1 expression. SCID mice implanted with GA0518 were randomly assigned to receive either control or LD2 treatment (20 mg/kg, daily, 5 times/ week for 2 weeks by intraperitoneal injection). Tumor volumes were observed and measured throughout treatment, and tumor weights and mouse body weights were measured at the end of treatment. The tumor volumes and tumor weights of the LD2-treated group showed a significant reduction compared with those of the control group ( Fig. 6G-I). In contrast, the mouse body weights did not differ significantly between the 2 groups (Fig. 6J), indicating that LD2 may be a relatively safe drug. Furthermore, we evaluated expression of GRK3, YAP1, and SOX9 and cell proliferation marker Ki67 in the tumor tissues of mice treated with or without LD2 using IF. All these markers were significantly reduced by LD2 treatment (Fig. 6K). These results confirmed that GRK3 inhibition by LD2 efficiently suppressed GAC tumor growth in vivo.
To further demonstrate the effect of GRK3 in the PCrelated mouse metastatic model (the tumor microenvironment is somewhat similar to that of metastatic human GAC), we injected MKN45 GFP control (NC) cells and GRK3 OE cells intraperitoneally in SCID mice. Our data showed that GRK3 OE in MKN45 cells significantly promoted PC measured by bioluminescence weekly, while LD2 treatment significantly reversed PC in GRK3 OE group (Supplemental Fig. 9) indicating that CRK3 is critical for peritoneal metastases and LD2 suppressed PC induced by GRK3 OE.

Discussion
Critical knowledge in new regulators and therapeutic targets for GAC patients is urgently needed. Through shRNA and cDNA functional screening of human kinases, we previously discovered that GRK3 is an essential kinase for prostate cancer progression. We found that GRK3 is significantly higher in GAC tumor tissues verse normal from TCGA dataset indicating its potential role in GAC. However, its expression and functions in GAC progression and metastases remain unclear. In the present study, for the first time, we showed that GRK3 expression was upregulated in GAC tissues and that higher GRK3 expression was significantly associated with higher TNM stage, more lymph node metastases, diffuse phenotype, and shorter survival. Importantly, we identified a novel GRK3 inhibitor LD2, which demonstrated strong anti-tumor activity in vitro and in vivo in the PDX model. Inhibition of GRK3, either pharmacologically or by genetic KD, in patientderived PC cells resulted in decreased cell proliferation, invasion, colony formation, and CSC attributes, while overexpression of GRK3 did the opposite. All these results suggest that GRK3 plays an oncogenic role in GAC. Moreover, we discovered a positive association between GRK3 and YAP1, and GRK3 upregulation in GAC cells increased YAP1 and its downstream targets, while GRK3 inhibition led to decreased in YAP1 and its downstream targets (Fig. 6L). Altogether, our study indicates that GRK3 is a poor prognosticator and a novel therapeutic target against GAC. Targeting GRK3 with LD2 may diminish aggressive phenotypes of GAC.
The clinical significance of GRK3 expression in different tumor types varies drastically and seems contextdependent. To our knowledge, our study is the first to report GRK3 expression in a large cohort of GAC patients, consisting of 393 pairs of the primary tumor and non-tumor adjacent tissues, as well as PC cells from metastatic GAC patients. By IHC and immunofluorescent staining, we found that GRK3 expression was higher in tumor tissues than in adjacent normal tissues and even higher in PC cells. Further analysis revealed the Fig. 5 GRK3 positively regulated YAP1 targets and GRK3 inhibitor reduced CSC attributes and aggressive phenotypes of GAC cells. A SOX9, Birc5, Cyr61 and CTGF were detected by qPCR in KATO III cells with or without GRK3 OE; **P < 0.001. B SOX9, Birc5, Cyr61 and CTGF were detected by qPCR in MKN45 cells with or without GRK3 OE; ***P < 0.0001.C SOX9 expression was detected by Western blot (upper panel) or co-IF in both KATO III and MKN45 cells with or without GRK3 cDNA overexpression (Bottom panel). Scale bar: 50µm. D The level of YAP1 and its targets SOX9, Cyr61 were detected using qPCR in GA0518 and GA0804 patient derived cells treated with or without LD2 at dosage indicated. ****P < 0.0001. E Level of SOX9 was determined by Western blot (upper panel) and co-IF (bottom panel) in GA0518 cells and treated with LD2 at indicated dosage. Scale bar: 50µm. F Tumor sphere formation was detected in MKN45 and KATO III cells with or without overexpression of GRK3 and then treated with LD2 at indicated dosage. *P < 0.01, **P < 0.001, ***P < 0.0001. G Representative image (left) and quantification (right) of ALDH1A1 + population was labeled using the ALDEFLUOR kit in KATO III cells with GRK3 OE compared to the control (n = 3 biological replicates). P = 0.000424. H Representative image (left) and quantification (right) of ALDH1A1 + population was labeled using the ALDEFLUOR kit in GA0518 cells with LD2 treatment at 0.5µM for 48 h (n = 3 biological replicates). P = 0.000741 overexpression of GRK3 in GACs of patients with diffuse type, lymph node metastases, and those with higher tumor stage. These results are consistent with findings reported in colon cancer and prostate cancer [7,12], although in breast cancer, decreased GRK3 expression correlates with worse phenotype and liver metastases [27]. Overall, our functional studies using pharmacologic and genetic modulations of GRK3 revealed that GRK3 plays an important oncogenic role in GAC, as the inhibition of GRK3 reduced GAC's aggressive phenotypes, including cell proliferation, tumor sphere formation, and invasion.
Mounting evidence suggests that GPCR regulates cell proliferation, stemness, self-renewal, and survival by modulating core cascades related to these processes, including small GTPases, MAPK, and PI3K/AKT/ mTOR pathways, as well as Wnt/β-catenin or YAP/TAZdependent transcription programs [28][29][30][31]. Seven GPCR kinases (GRKs) in human genome critically regulate signaling pathways and functions of nearly 1,000 GPCRs [28,32]. GRK3 is one of the 7 ubiquitously expressed GRKs that regulate the functionality of both G proteincoupled receptors (GPCR) and growth factor receptors and to directly control cytosolic, cytoskeletal or nuclear signaling components of pathways relevant for these processes [28]. GRK3 has been reported to phosphorylate the agonist-occupied form of β-adrenergic receptors and several other GPCRs to regulate their signaling either positively or negatively [33][34][35][36]. GRK3 may promote cancer development by facilitating GPCR-mediated oncogenic functions through phosphorylation of some GPCRs.
To investigate mechanisms by which GRK3 exerts its oncogenic functions in GAC, we focused on GRK3-YAP1 axis, due to the importance of Hippo/YAP1 signaling in GAC, such as conferring CSC attributes and promoting metastases. Several lines of evidence from our study support this postulation. First, there was a strong positive correlation between GRK3 and YAP1 upon IHC staining in TMAs, which was confirmed using qPCR in a separate patient cohort as well as TCGA dataset, including primary tumors and PC cells. Second, genetic or pharmacologic inhibition of GRK3 dramatically decreased YAP1 expression and its targets SOX9, Birc5, CYR61, and CTGF, accompanied by a reduction in aggressive phenotypes and CSC attributes in GAC cell lines and patient-derived cells. Conversely, GRK3 overexpression in both KATO III and MKN45 induced YAP1 targets and increased aggressive phenotypes of GAC cells. More detailed mechanisms on how GRK3 regulates the activation of YAP1 signaling remains a subject of active investigation in our laboratories.
As kinases have proven to be suitable drug targets for cancer therapeutics [34], to translate our findings into the clinic, it is necessary to identify an effective GRK3 kinase inhibitor. By collaborating with Dr. Kevin Dalby's laboratory at The University of Texas at Austin, we screened a compound library to identify GRK3 kinase inhibitors using recombinant GRK3 protein and Z-LYTE in vitro kinase assay. We identified LD2, which potently inhibited GRK3 kinase activity (Figs. 3B & 3C). Of note, GRK2, the GRK3's closest sibling kinase, is much less potently inhibited by LD2 than GRK3. Interestingly, the structure of LD2 (Fig. 3A) coincides with that of PF-477736, a CHK1 inhibitor [35,36]. Importantly, we found that LD2 suppresses GRK3-mediated oncogenic activity in GAC cell lines and patient PC cells, whose inhibitory effects are less dramatic in GRK3-low cancer cells. LD2, a novel GRK3 inhibitor, actively reduced tumor cell growth, invasion, and tumor sphere formation in vitro in multiple GAC cell models and in vivo in a GAC PDX model with high GRK3 and YAP1 expression. The evidences here provide a strong rationale for targeting GRK3 with LD2 and its derived compounds in GAC patients. Details on the mechanism action of LD2 and the relationship between inhibitions of CHK1 and GRK3 kinase activities warrant further investigation.
(See figure on next page.) Fig. 6 GRK3 promoted tumor growth, while inhibition of GRK3 by LD2 significantly suppressed GAC growth in xenografts. A Representative tumor images and tumor volume were shown in MKN45 GRK3 overexpression (OE) vs negative control group (NC) over three-week period after implantation. B-D Tumor burden, tumor weights and tumor volumes were shown at end point. E GRK3 level was detected by qPCR in treated tumor tissues compared to NC group. ***P < 0.001. F Co-immunofluorescent staining of GRK3 and SOX9 was performed in tumors of GRK3 OE group vs NC group. Representative images were shown. Scale bar: 25µm. G Representative GA0518 xenograft tumors in LD2 treatment group vs control were shown after 2-week treatment. H & I tumor weights (H) and tumor volumes (I) were significantly reduced in the LD2-treated group compared with that of the control group. *P < 0.01; **P < 0.001. J Mouse body weight was measured in the indicated group. K Immunofluorescent staining of GRK3, YAP1, SOX9 and proliferation marker Ki67 was performed in LD2 treated tumor tissues compared with control tumor tissues. Images were captured by confocal microscopy. Scale bar: 25µm. L Diagram demonstrates the link between GRK3 and YAP1 and its downstream targets. Novel GRK3 inhibitor LD2 can curtail tumor cell progression and metastases mediated by the GRK3-YAP1 axis

Conclusion
In conclusion, we uncovered, for the first time to our knowledge, that GRK3 is over-expressed in primary and metastatic GAC tumors compared with adjacent normal tissues, where its overexpression is significantly associated with poor prognosis and shorter survival. Functional studies using genetic and pharmacologic approaches in the various GAC cell models and patientderived cells confirm that GRK3 plays an oncogenic role. Furthermore, we have discovered a GRK3 inhibitor, LD2, that suppresses GAC aggressive attributes and tumor growth in vitro and in vivo. Thus, GRK3 could serve as a prognostic biomarker and a therapeutic target for advanced GAC patients who are in desperate need of better treatments.