Up‐regulated acylglycerol kinase (AGK) expression associates with gastric cancer progression through the formation of a novel YAP1‐AGK–positive loop

Abstract Acylglycerol kinase (AGK) uses adenosine triphosphate (ATP) and acylglycerol to generate adenosine diphosphate (ADP) and acyl‐sn‐glycerol 3‐phosphate in cells. Recent evidence has demonstrated that dysregulated AGK expression is associated with the development of various human cancers. This study investigated the effects of AGK on gastric cancer cell proliferation and carcinogenesis and explored the underlying molecular events. AGK expression was up‐regulated in gastric cancer and was associated with poor prognosis in gastric cancer patients. AGK overexpression increased gastric cancer proliferation, invasion capacity and the expression of the epithelial‐mesenchymal transition markers in vitro. Conversely, the knockdown of AGK expression reduced gastric cancer cell proliferation in vitro and in nude mouse tumour cell xenografts. Importantly, AGK expression was associated with the YAP1 expression in gastric cancer cells and tissues. YAP1 expression also transcriptionally induced AGK expression through the binding of TEAD to the AGK gene promoter. However, AGK expression inhibited the activation of the Hippo pathway proteins and induced YAP1 nuclear localization to enhance the transcription activity of YAP1/TEADs. In conclusion, the study demonstrates that AGK is not only a novel target of the Hippo‐YAP1 pathway, but that it also positively regulates YAP1 expression, thus forming a YAP1‐AGK–positive feedback loop.

of novel prognostic markers and therapeutic strategies to control this deadly disease. Previous studies have demonstrated that the Hippo pathway plays a critical role in the regulation of organ size during embryo development and tissue homeostasis in adults. On the molecular level, for example, when the Hippo pathway is activated, macrophage-stimulating protein 1/2 (MST1/2) will phosphorylate large tumour suppressor kinase 1/2 (LATS1/2) and the latter will subsequently phosphorylate and repress the activity of Yes-associated protein 1 (YAP1) and its homolog transcriptional co-activator TAZ (also known as WWTR1). [4][5][6] Thereafter, YAP1 that is phosphorylated at S127 or TAZ that is phosphorylated at S89 will bind to the 14-3-3 protein, leading to YAP1 and TAZ retention in the cell cytoplasm; otherwise, YAP1 and TAZ proteins will translocate into the nucleus where they activate transcription and the expression of their downstream genes (like CTGF, CYR-61, FOXM1 and CDX2) to facilitate their biological functions in cells, such as cell proliferation and migration. [7][8][9] Accumulating evidence suggests that dysregulation of the Hippo pathway signalling is associated with the development, progression and metastasis of different human cancers. 10,11 Thus, alteration of YAP1 expression and activity has also been shown to be associated with cancer development. 12,13 In gastric cancer, YAP1 expression contributes to poor patient survival. 14 YAP1 acts as an oncogene or possesses an oncogenic effect in gastric cancer 15 and is able to promote gastric cancer cell survival and migration. 16 Although a few regulatory factors that can act on Hippo-YAP1 pathway have been uncovered, the mechanism of Hippo inactivation and YAP1 relevant transcriptional targets in GC are still remained incompletely understood.
Acylglycerol kinase (AGK), acting as a lipid kinase, functions to phosphorylate monoacylglycerol and diacylglycerol to form lysophosphatidic acid (LPA) and phosphatidic acid (PA), 17 resulting in the activation of the downstream signalling. [17][18][19] Recently, AGK was shown to be a cancer-related protein that is overexpressed in various human cancers, such as prostate cancer, 20 hepatocellular carcinoma, 21 breast cancer 22 and oesophageal squamous cell carcinoma (ESCC). 23 The level of AGK expression is significantly associated with the Gleason scores and capsular invasion of prostate cancer, 20 the angiogenesis and tumour cell survival of hepatocellular cancer, 21 and the sustained constitutive JAK2/STAT3 activation in oesophageal squamous cell carcinoma. 23 In human carcinogenesis, both the Hippo-YAP1 and AGK pathways co-ordinately play a role, although the precise protein-protein interactions and molecular pathways require further investigation. Therefore, in this study, we investigated the effects of AGK on gastric cancer cell proliferation and carcinogenesis and the underlying molecular events. We expected to provide novel information regarding the role of the AGK and Hippo-YAP1 pathways in the development of gastric cancer and determine whether they should be further evaluated as biomarkers for the early detection and prediction of prognosis in gastric cancer and explored as therapeutic targets for the treatment of gastric cancer.

| Tissue samples, cell lines and culture
In this study, we collected 120 gastric cancer tissue samples in the form of paraffin blocks from the Department of Pathology, The First Affiliated Hospital to Nanchang University (Nanchang, China). The samples were taken from patients that were hospitalized between January 2009 and December 2012 and were histologically diagnosed with gastric cancer. None of the included patients received any presurgery chemotherapy. The clinicopathological data from each patient were collected from their medical history (Table S1). Tumour stages were classified according to the 2010 criteria of the American Joint Committee on Cancer. 24 In addition, we obtained 12 fresh gastric cancer and paired non-cancerous mucosal tissues from the surgery room, which were immediately snap-frozen in liquid nitrogen and stored at −80°C until use. This study was approved by the Ethics Committee  and mouse monoclonal anti-GAPDH (GTX627408; GeneTex, 1:1000).

| Protein extraction and Western blot
Antibodies were used according to the manufacturer's protocols.

| Immunohistochemistry
Immunostaining of YAP1 and AGK proteins in paraffin sections of the gastric cancer tissues followed a protocol that was described in a previous study. 25 The staining results were evaluated by two pathologists, based on the proportion of positively stained cells and the intensity of staining, that is % of staining was scored as 0 (0%), 1 (0%-10%), 2 (10%-50%) and 3 (50%-100%), while the intensity of staining was as: 0 (negative), 1 (weak), 2 (moderate) and 3 (strong).
These two scores were then multiplied to form a staining index. In cases where the staining index was <4, it was classed as low expression, whereas in cases where the staining index was ≥4, it was classed as high expression of YAP1 or AGK.

| Plasmid construction and cell transfection
To knock down the gene expression, we designed four different siRNAs, and a negative control, that is YAP1 siR-1, 5′-CUG

| Immunofluorescence
Cells were seeded into a 6-well plate and grown overnight to reach appropriate confluency and then transiently transfected with siAGK, Flag-AGK plasmid and their corresponding negative controls for 48-72 hours. At the end of each experiment, cells were washed with ice-cold phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde for 15 minutes, and then permeabilized in 0.2% Triton X-100 (Roche Diagnostics Co., Indianapolis, IN, USA) for 15 minutes and blocked in 2% bovine serum albumin (BSA)/PBS at the room temperature for 60 minutes. The cells were subsequently incubated with a primary antibody, diluted with 0.1% BSA at 4°C overnight, and on the next day, the cells were washed three times with PBS and further incubated with a fluorescent dye-labelled secondary antibody at room temperature for 45 minutes in the dark, and reviewed under a fluorescence microscope (Nikon, Tokyo, Japan).

| Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR)
Total cellular RNA was isolated from cells using a TRIzol reagent (Invitrogen) and reversely transcribed into cDNA using the

| Cell viability CCK-8 assay
Gastric cancer cells were seeded into 96-well plates and transfected with different genes for 48 hours. Cells were then subjected to a cell viability assay, comprising up to 120-hour incubation. At the end of each experiment, the cell culture was combined with 10 μl of CCK-8 solution (TransGen Biotech), further incubated for 4 hours, and the optical density value was measured by using a microplate reader (Thermo Scientific) at the absorbance at 450 nm.
The experiments were performed in triplicate and repeated at least three times.

| Tumour cell colony formation assay
Gastric cancer cells were seeded into 6-cm dishes and transfected with different genes for 24 hours. Cells were then subjected to a colony formation assay, that is cells were trypsinized and re-seeded into 6-cm plates with a density of 1000 cells per well in triplicate and cultured for 14 days, throughout which time the medium was exchanged every three days. At the end of each experiment, cells were fixed with methanol and stained with 0.1% crystal violet solution and the numbers of cell colonies were counted.

| Nude mouse tumour cell xenograft assay
The animal protocol of this study was approved by the Institutional Animal Care and Use Committee (IACUC) of The First Affiliated Hospital to Nanchang University (Nanchang, China). In this study, BALB/c nude mice at four weeks of age were purchased from the Center of Experimental Animal of Guangzhou University of Chinese Medicine (Guangzhou, China) and randomly assigned to two groups (n = 6). Scramble short hairpin RNAs or short hairpin RNAs targeting AGK were subcloned into the lentiviral expression vector, GV248 (Genepharma). Animals were then subcutaneously injected with 2 × 10 6 of HGC-27 cells that were stably expressing LV-shAGK or LV-scramble shRNA. Growth of the tumour cell xenografts was monitored daily and two-dimensional measurements were made using electronic digital calipers (Thermo Scientific) at the indicated period of time ( Figure 2E). Tumour volume was calculated by using the formula of 3.14/6 × L × W 2 (L = tumour length, W = tumour width). After 31 days, the mice were killed by euthanasia and tumour xenografts were harvested and weighed.

| Luciferase reporter assay
HEK-293T cells were seeded into 24-well plates at a density of 2 × 10 6 /well in triplicate, grown overnight and transfected with the indicated plasmids using a Lipofectamine 2000 reagent (Invitrogen), according to the manufacturer's instructions for 48 hours. After that, the luciferase activity was measured by using the Dual-Luciferase Reporter Assay System (Promega), according to the manufacturer's protocol.

| Statistical analysis
The data were statistically analysed using the SPSS

| Up-regulation of AGK protein in gastric cancer tissues and cell lines, and the association with clinicopathological features of patients
In this study, we first assessed AGK expression using online database data and found that AGK was not significantly amplified in gastric cancer on the TCGA database (http://www.cbiop ortal.org/) data ( Figure 1A). However, the TCGA-STAD database data showed that the level of AGK mRNA was dramatically higher in gastric cancer tissue than in normal tissues ( Figure 1B). Our own Western blot and immunohistochemistry data show that the expression of the AGK In the current study, the AGK expression was analysed in the tissue samples of 120 cases of gastric cancer using immunohistochemistry ( Figure 1G). We then associated AGK expression with clinicopathological data from patients and found that AGK expression was significantly associated with histological differentiation (P = .009), but not with other clinicopathological data (Table S1).
The Kaplan-Meier curves and the log-rank analysis revealed that the overall survival was lower in patients with high AGK-expressing tumours than in the patients with low AGK-expressing tumours l (53.8 ± 4.2 months vs 56.7 ± 3.2 months; P = .045; Figure 1H).
Furthermore, bioinformatic analysis of the online KM plotter database data (kmplot.com/analy sis/index.php?p=servi ce&cance r=gastric) supported our findings ( Figure 1I).  Conversely, overexpressing AGK led to opposite results in BGC-823 cells. C, Immunofluorescence results for EMT markers after GC cells were transfected with AGK siRNA or cDNA. The data are summarized as the mean ± SD of three independent experiments. *P < .05  Figure 4A). The Pearson chi-square test result shows that the nuclear YAP1 expression was highly associated with cytoplasmic AGK expression ( Figure 4B). In addition, the TCGA database data (http://gepia.cance r-pku.cn/detail.php?click tag=corre lation) further confirmed this finding, showing that AGK expression was associated with YAP1 expression in gastric cancer tissues ( Figure 4C).

| Association of YAP1 and AGK expression in gastric cancer ex vivo and in vitro
Additionally, we found that knockdown of AGK decreased the level of CTGF, and this effect was reversed by overexpression of YAP1.
We also noticed that transfection of YAP1 increases AGK protein level ( Figure 4D). This Western blot result and the observation that YAP1 and AGK were up-regulated and highly associated in GC cell lines and tissues, raising that there might be a reciprocal interplay between YAP1 and AGK.

| YAP1 transcriptional up-regulation of AGK expression in gastric cancer cells
To further confirm the role of YAP1 in the regulation of AGK in gastric cancer cells, we associated AGK expression with other YAP1related proteins in gastric cancer tissue samples and found that AGK expression was also associated with the expression of TEAD1, TEAD2 and TEAD4 ( Figure S1A). Thus, we knocked down or overexpressed YAP1 using YAP1 siRNA and cDNA, respectively, and found that knockdown or overexpression of YAP1 in BGC-823 and HGC-27 cells changed the expression of AGK and CTGF proteins, respectively ( Figure 5A). Furthermore, we treated gastric cancer cells with verteporfin or metformin, the YAP inhibitors, 28  for 48 h and analysed for YAP1, AGK and CTGF expression. C, qRT-PCR results indicated that after knocking down YAP1, AGK along with Hippo-YAP1 target gene mRNAs, such as for CTGF and CYR61, were decreased in HGC-27 cells, while overexpressing YAP1 increased AGK, CTGF and CYR61 mRNA levels in BGC-823 cells. D, AGK promoter activity was detected in HEK-293T cells that transfected with YAP1 cDNA, YAP-S94A or the combination of YAP1 and TEAD1-4 cDNA. E, Illustration of luciferase reporter constructs (wild-type and mutated-type). Mutant luciferase constructs were generated by site-directed mutagenesis of TRE1 or TRE2 alone or in combination. F, HEK293T cells were grown and cotransfected with the indicated plasmids and then subjected to luciferase reporter assay. The mutated TRE1 in the AGK promoter was sufficient to abolish AGK promoter activity induced by YAP/TEAD4 cDNA. G, Design of the AGK promoter primers for ChIP assays. H, Chromatin from HGC-27 was pulled down using a YAP1 antibody or IgG. PCR amplification using TRE1 or TRE2 primers was then performed. A 292 bp PCR product containing TRE1 in the AGK promoter was amplified from the chromatin DNA that was pulled down by the YAP antibody, while the input chromatin is shown as a positive control for ChIP assay family member (TEAD1-4) further enhanced the luciferase activity of the AGK promoter compared with the results when the cells were transfected with YAP1 only ( Figure 5D). The YAP1 + TEAD4 group showed the strongest induction of luciferase activity of all of the groups that were tested ( Figure 5D). This is consistent with the results of a previous study that identified TEAD4 as the major YAP1 interacting TEAD family member. 29 Furthermore, we also assessed whether the YAP1/TEADs interaction is essential for AGK transcription by generating a mutated YAP-S94A that lacks TEAD-binding capacity. Our data show that this mutated YAP-S94A failed to activate the luciferase activity of the AGK promoter ( Figure 5D). Taken together, these results suggest that AGK could be a transcriptional target of YAP1/TEADs.
To further map the TEAD response element (TRE) that is responsible for TEAD binding and AGK promoter activation, we found two potential TREs (TRE1 and TRE2) and mutated them individually ( Figure 5E). The luciferase reporter assay used these mutated constructs after cotransfection with YAP1 and TEAD4 and showed that TRE1 mutation was sufficient to abolish AGK promoter activity ( Figure 5F). Furthermore, we performed a chromatin immunoprecipitation assay to further validate the binding of TEAD to the AGK promoter and found that YAP1/TEADs was able to bind to TRE1 in the AGK promoter, but not TRE2 to induce AGK transcription ( Figure 5G,H, and Figure S1B).

| AGK inhibits the Hippo-YAP1 pathway and upregulates TEAD transcriptional activity
We then assessed the interaction of AGK and Hippo-YAP1 pathway in gastric cancer cells and found that AGK overexpression was able to up-regulate level of YAP1 and CTGF proteins, but lead to a down-regulation of the level of LATS1/2 and p-YAP without changing in MST1/2 levels ( Figure 6A). Furthermore, knockdown of AGK expression had the reverse results ( Figure 6B). Given the association of AGK with the Hippo-YAP1 pathway, we speculated whether the YAP1 level was determined by the AGK expression.
Thus, we overexpressed and knocked down the expression level of AGK in BGC-823 cells and found that AGK overexpression Furthermore, AGK overexpression increased the level of these mRNAs ( Figure 6E). Taken together, our data indicate that AGK is able to up-regulate the expression of the Hippo-YAP1 pathway downstream genes through the enhancement of the transcriptional activity of TEADs. Our findings suggest that AGK is not only a novel target of the Hippo-YAP1/TEADs pathway but also a repressor of the Hippo pathway; thus, acting as a stimulator of the transcription activity of YAP1/TEADs to form a positive feedback circuit in gastric cancer cells ( Figure 6F).

| D ISCUSS I ON
The abnormal expression or dysregulation of multiple signalling pathway genes contributes to gastric cancer development and progression. 31  to the mutated AGK gene promoter. These binding sites were localized at −1757 to −1748 nt (5′ATGGTATTTG-3′) and −698 to −689 (5′-ACAGAATGTA-3′) of the AGK gene. Our Western blot analysis revealed that up-regulated AGK expression led to a reduction of LATS1/2 and phosphorylated YAP at S127, although it rescued the level of total YAP1 protein, whereas knockdown of AGK expression had the opposite effects. These data indicate an interaction of AGK with YAP1/TEADs proteins in the regulation of gastric cancer cell proliferation and tumour progression. Furthermore, our immunofluorescence data localized YAP1 proteins in gastric cancer cells after knockdown or overexpression of AGK, that is there was an increase in the expression of nuclear YAP1 protein in AGK-overexpressing cells, indicating that the YAP1 protein is activated in gastric cancer cells after AGK overexpression. Indeed, our luciferase reporter assay results show that AGK was able to induce the transcription activity of YAP1/TEADs genes. Previous studies have reported that LPA is a small molecular YAP1 activator and is able to induce LATS expression. 35,36 Intracellular LPA is generated by AGK phosphorylation of monoacylglycerol; however, our current data did not observe any restoration of YAP or LATS expression in the cells in which AGK was knocked down following treatment with LPA ( Figure S1C), indicating that AGK activity might function differently from YAP1 expression.
Moreover, our current data show that AGK overexpression reduced the half-life of LATS1/2 proteins (Figure S1D), indicating a post-transcriptional regulation of LATS1/2 protein in the Hippo-YAP1-related gene signalling. Now, our group is further investigating the regulation and functions of AGK and YAP1 gene pathways in gastric cancer development and progression. In conclusion, our current study provides the first evidence that AGK expression is up-regulated in gastric cancer cells and tissues, and that the up-regulation of AGK is associated with poor overall survival of gastric cancer patients.
We also demonstrated that AGK is a novel transcription target gene of YAP1/TEADs and is able to induce YAP1/TEADs transcription activity through the inactivation of the Hippo pathway to provide a positive feedback loop of YAP1 interaction with AGK in gastric cancer cells. Additionally, our study also delineated the feedback regulation between AGK and the Hippo pathway, which will contribute to a better understanding of the molecular mechanisms of AGK functions in gastric cancer. Future studies should assess whether the targeting of AGK expression or activity could be a therapeutic target for the control of gastric cancer progression.