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‐
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.
Materials and methods
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 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.
Indirect immunofluorescence
PC cells were used to produce cell blocks for slides, and some paired primary tumor tissues were subjected to indirect immunofluorescence staining with anti-GRK3 (1:100), anti-SOX9 (1:2000), anti-CTGF (1:100), anti-Cyr61 (1:100), and anti-Survivin (1:100) then labeling with Alexa Fluor 555 (for GRK3, Cyr61, and Survivin) and Alex Fluor 488 (for SOX9 and CTGF) as described previously [
17,
18]. Fluorescence was observed on a confocal microscope (FluoView FV500; Olympus America, Melville, NY) and analyzed by CellQuest Pro software (BD Biosciences, Franklin Lakes, NJ).
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/designcrispr.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.
GRK3 cDNA overexpression, constitutive shRNA silencing and inducible shRNA silencing were achieved by viral transduction and antibiotics selection, similarly to what we previously described [
7]. Briefly, GRK3 wild type (WT) or kinase-dead mutant cDNA were in pJP1653 retroviral vector (blasticidin selection). Constitutive shGRK3 and control scramble shRNA were in pLKO-puro lentiviral vector (puromycine selection). Doxycycline-inducible shGRK3 and control scramble shRNA were in pCW39-neo vector (an inducible vector similar to Addgene's pLKO-Tet-On, G418 selection). Doxycycline at 0.2–0.5 μg/ml for 3–4 days was employed to inducibly expressing shGRK3. The rescue experiment in KATO III and MKN45 GRK3 OE cells by KD YAP1 using LentiCRISPR/CAS9 technology. Three pair of YAP1 guide RNA followed German Cancer Research Center (
http://www.e-crisp.org/E-CRISP/designcrispr.html) guide RNA (gRNAs) design guidelines (Supplementary Table
1). The designed pairs of gRNA were T4 ligated to the pLentiCRISPR V2mOrange (V2mO, # 140,206, Addgene, Cambridge, MA). Clones are sequencing verified. Then, these KD virus from 293 T cells were transduced into the KATOIII and MKN45 GRK3 OE cells to generate YAP1 KD cells in GRK3 OE in both KATOIII and MKN45 cells to be able to perform the rescue 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.
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 × 106 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).
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 patient-derived 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 context-dependent. 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 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/TAZ-dependent transcription programs [
28‐
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 protein-coupled 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‐
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.
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 patient-derived 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.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.