Preclinical evaluation of exemestane as a novel chemotherapy for gastric cancer

Abstract CYP19A1/aromatase (Ar) is a prognostic biomarker of gastric cancer (GCa). Ar is a critical enzyme for converting androstenedione to oestradiol in the steroidogenesis cascade. For decades, Ar has been targeted with Ar inhibitors (ARIs) in gynaecologic malignancies; however, it is unexplored in GCa. A single‐cohort tissue microarray examination was conducted to study the association between Ar expression and disease outcome in Asian patients with GCa. The results revealed that Ar was a prognostic promoter. Bioinformatics analyses conducted on a Caucasian‐based cDNA microarray databank showed Ar to be positively associated with GCa prognosis for multiple clinical modalities, including surgery, 5‐Fluorouracil (5‐FU) for adjuvant chemotherapy, or HER2 positivity. These findings imply that targeting Ar expression exhibits a potential for fulfilling unmet medical needs. Hence, Ar‐targeting compounds were tested, and the results showed that exemestane exhibited superior cancer‐suppressing efficacy to other ARIs. In addition, exemestane down‐regulated Ar expression. Ablating Ar abundance with short hairpin (sh)Ar could also suppress GCa cell growth, and adding 5‐FU could facilitate this effect. Notably, adding oestradiol could not prevent exemestane or shAr effects, implicating a nonenzymatic mechanism of Ar in cancer growth. Regarding translational research, treatment with exemestane alone exhibited tumour suppression efficacy in a dose‐dependent manner. Combining subminimal doses of 5‐FU and exemestane exerted an excellent tumour suppression effect without influencing bodyweight. This study validated the therapeutic potentials of exemestane in GCa. Combination of metronomic 5‐FU and exemestane for GCa therapy is recommended.


| INTRODUC TI ON
Gastric cancer (GCa) is the third leading cause of cancer-related mortality in the world. 1 The incidence of GCa has been reported to vary worldwide, 2,3 and GCa has a poor prognosis with an only ≤10% 5-year survival rate. 4 Most patients are diagnosed at an advanced stage, or they rapidly experience relapse within 12 months after surgery. [4][5][6] Resection is the first-of-choice treatment modality; nevertheless, it is associated with a high recurrence rate. 6 Conversely, chemotherapy is often effective in patients with early-stage GCa; however, poor prognosis is still presented in patients with advanced GCa. 2,4,7 Therefore, there is high clinical demand for new adjuvant chemotherapy for GCa. 2 Gastrectomy is the main therapeutic modality for GCa.
Nevertheless, postsurgical recurrence is often observed in advanced disease. 8 Therefore, adjuvant chemotherapy is used for secondary prevention. 4 Among various chemoagents, 5-Fluorouracil (5-FU)based adjuvant therapy drugs are commonly used. 9 Therefore, regarding new drug development, the therapeutic outcome of 5-FU is often used as the baseline for comparison in patients with GCa. 7 Studies have shown that 15% of patients with GCa are HER2 positive, 10,11 and such patients can be treated with a HER2 inhibitor as alternative once 5-FU fails.
A study revealed that lipoprotein protein/receptor-route-mediated cholesterol import and the resulting steroidogenesis play crucial roles in GCa progression. 12 CYP19A1 (cytochrome P450 family 19 subfamily A member 1; also named aromatase, Ar) is the key enzyme catalysing the conversion of androstenedione or testosterone to oestradiol or oestrone. Several studies have shown that breast cancer progression is also associated with Ar, whereas Ar inhibitors (ARIs) were also implemented in a therapeutic regimen. 1,13 Whether ARIs can be used for GCa therapy is an intriguing question. Several cohort studies have reported that Ar expression was higher in GCa tumours than it was in normal parts. 12,14,15 This information has motivated the evaluation of the clinical value of Ar.
The mechanism of action (MOA) of ARIs involves inhibiting Arenzymatic function 16 ; this therefore reduces oestrogen levels in organisms. 17,18 ARIs are commonly used in the clinical treatment of breast cancer. 19 They constitute an adjuvant hormonal therapy for patients with oestrogen receptor (ER) or for postmenopausal patients to reduce breast cancer risk. 20 ARIs can be classified as type I (ARI-I; eg, anastrozole and letrozole [nonsteroidal]) and type II (ARI-II; eg exemestane [steroidal]). 18,21 However, these two types of ARIs act in different inhibitory modes against Ar ARI-I interacts with the catalytic site of Ar by inhibiting anastrozole conversion. The mode of action of ARI-I entails reversibly blocking the interaction between anastrozole and Ar, thus leading to elevated Ar levels in cells. By contrast, ARI-II irreversibly inhibits Ar by forming a covalent bond with the catalytic site of Ar Thus, ARI-II irreversibly inactivates Ar function, leading to diminished Ar levels in cells. 22 Additionally, the different modes of action of ARI-I and ARI-II trigger diverse molecular mechanisms, although they target the same proteins. 23 In this study, we used a bioinformatics approach to explore the suitability of Ar as a targeting agent for treating GCa, thus meeting clinical needs. We also tested the value of targeting Ar with ARIs in GCa preclinical models and provide new insight into ablating Ar with a non-irreversible inhibitor in GCa treatment. and 2014. Patient characteristics and clinical outcome were followed until death, censorship or loss to follow-up. Gastric tumour tissue cores were collected from each patient and used to construct a tissue microarray (TMA). 24 The clinical parameters and overall survival (OS) data were obtained from patients' medical records followed up for 5 years.

| Kaplan-Meier plotter for cancer survival analysis
To analyse the association of survival with gene expression, a webbased Kaplan-Meier plotter (http://kmplot.com/analy sis/index. php?p=servi ce&cance r=gastric) was used and a log-rank test was used to assess the differences between patient groups stratified according to the median of gene expression. A P-value of <.05 was considered statistically significant. subminimal doses of 5-FU and exemestane exerted an excellent tumour suppression effect without influencing bodyweight. This study validated the therapeutic potentials of exemestane in GCa. Combination of metronomic 5-FU and exemestane for GCa therapy is recommended.

K E Y W O R D S
aromatase, exemestane, gastric cancer

| Statistical analysis
Statistical analyses were performed using Student's t test. All experiments were repeated at least three times, and P-value <.05 was considered to be statistically significant.
Other materials and methods (including reagents, cytotoxic assay, IC.50 measurement, colony-forming assay, gene expression measurements, knockdown of Ar in the cells and cancer cells xenograft assay) are in the online supplemental materials .

| Ar expression is the GCa prognosis biomarker
To assess the effect of Ar expression in clinical settings, we implemented two strategies to evaluate the roles of Ar in human GCa progression: one of the strategies entailed conducting a single-hospital cohort TMA immunohistochemistry study on an Asian population (Tables 1 and 2); the other entailed using a Kaplan-Meier plotter for survival analysis to determine the association of gene expression with Caucasian GCa prognosis ( Figure 1). In the single-cohort TMA study, men constituted a predominant portion of the study population (111 men vs 65 women). The Ar-positive staining can be found in men (14%) and women (22%). The positivity in normal parental was higher in women (67%) than it was in men (41%) ( Table 1; P = .0017).
The GCa death rate was higher (P = .0283) in women (46%) than it was in men (27%). Notably, the association of tumour Ar positivity with death rate was higher in men (10 of 16 high-expression patients

| Modulation of Ar expression suppresses GCa cell growth
As mentioned, two MOAs of ARIs were considered to test whether ARIs can be used for GCa therapy. Three ARIs (ARI-I: anastrozole and letrozole, reversible inhibitor; and ARI-II: exemestane, irreversible inhibitor) were introduced. Cytotoxicity (Figure 2A,2B) and colony formation capacity ( Figure 2C) were measured, and the results revealed that exemestane had excellent cytotoxic against GCa cells. By contrast, we did not observe a significant cytotoxic effect of anastrozole or letrozole on GCa cells, implying that ARIs have a different mode of action in GCa. In addition, exemestane could suppress Ar expression at the transcriptional level ( Figure 2D). The discrepancy in cytotoxic efficacy between ARI-I and ARI-II raised the question whether Ar expression but not enzymatic activity (concerting androgens to oestrogens) may be crucial for cytotoxic efficacy against GCa cells.
To test this hypothesis, we introduced gene-silencing technology with shRNA targeting Ar mRNA expression ( Figure 3A). We compared the colony-forming ability of shLuc cells with that of shAr infectants and found that the shAr infectants had a lower colonyforming ability than did the shLUC infectants. To examine whether Ar down-regulation can alter the sensitivity of 5-FU to GCa, we tested the response of 5-FU to shLuc and shAr infectants. The results are presented in Figure 3C, demonstrating an excellent synergistic effect in the combined treatment. To rule out the possibility of Ar-enzymatic-activity-mediated oestradiol (E2) production, we added E2 to both shLuc and shAr infectants to examine whether E2 Ar knockdown and 5-FU ( Figure 3D). In addition to the short-term cytotoxic effect of the combined treatment, we observed that the combined treatment had a long-term effect on colony formation in GCa cells. Specifically, as shown in Figure 3E,F, with a combination of Ar knockdown and 5-FU, the colony-forming ability was significantly reduced; adding E2 did not reverse the effect of the combined treatment, suggesting that 5-FU cytotoxicity may be enhanced through Ar-meditated signalling independent of oestrone conversion.
Overall, our findings provide proof-of-concept evidence that ablating Ar expression using exemestane or shRNA is potentially therapeutic.

| Preclinical evaluation of 5-FU and exemestane combination therapy
As shown in Figure 3D,E, a synergistic inhibitory effect was observed for the combination treatment of Ar knockdown and 5-FU.
Combining exemestane and 5-FU has high potential for GCa treat- to P = .007). Considering the effects of the treatments on the general wellness of the experimental mice, we divided tumour weight by bodyweight ( Figure 5C). We used it as the basis to compare within the treatment groups. We determined that combination treatment was the best scheme for therapy.
In sum, targeting Ar with exemestane (ARI-II) might be a new effective therapeutic approach for GCa. Single therapy or combination usage with 5-FU is worthwhile to be tested in clinical settings.

| D ISCUSS I ON
Gastric cancer is a complex malignancy because patients are usually diagnosed at an advanced stage and have a poor prognosis. 4 Since the early 1980s, fluorouracil-based chemotherapy is considered a standard treatment for GCa. 12 However, 5-FU is commonly associated with poor selectivity and systemic toxicity. 25 Therefore, seeking a more effective target is crucial for GCa therapeutics. In this study, we first discovered that Ar expression might be a new prognostic biomarker and an important gatekeeper in multiple treatment modalities for patients with GCa; thus, Ar has potential to serve as a targeting agent to develop medications to fulfil unmet medical

| Potential hazard of anti-ER therapy for GCa treatment
The contribution of female factors to GCa development is controversial. For example, a large-scale epidemiological survey indicated female factors, such as reproductive age, ovariectomy surgery, breastfeeding, pregnancy and contraceptive agents, were suggested that the oestrogenic signal suppresses GCa incidence. 26  in GCa therapy. However, anti-ERs (eg tamoxifen) may lead to the development of gastric, oesophageal and colorectal malignancies. [30][31][32] In the current study, we targeted the upstream region of oestrogen/ ER, oestrone synthesis, in cancer cells and demonstrated excellent tumour suppression efficacy.

| Possible mechanism of Ar silencing with exemestane
Exemestane inhibits Ar expression, which is considered an essential MOA of GCa cell suppression. We propose several mechanisms of Ar silencing. The first mechanism is transcriptional regulation; However, this hypothesis was not favoured in this study because ARI-I did not exert suppressive effects on Ar expression (data not shown). The third mechanism is off-target inhibition; for example, Ar activity could increase prostaglandin E2 (PGE2) binding to the G-protein-coupled PGE2 receptor to stimulate cyclic AMP production. 43,44 Whether the consequential PGE2 ablation with ARI-II inhibits cell growth requires further examination. The fourth mechanism is the epigenetic modification of Ar by ARI-II. For example, a potent Ar expression inhibition agent, namely LBH589, can selectively suppress the human Ar gene promoter I.3/II by reducing C/EBPδ levels. The decreased binding of C/EBPδ on Ar could increase the levels of acetyl-histones on the promoter I.3/ II, thus silencing Ar expressions. 45 Another highly possible mechanism is that ARI-II suppresses C/EBPδ to silence Ar transcription.

| CON CLUS ION
In this study, we observed that Ar is a crucial GCa prognostic biomarker.
Suppressing Ar expression by using ARI-II could be an excellent therapeutic strategy, particularly when ARI-II is used in combination with 5-FU. Additional pharmaceutical studies and human trials are encouraged.

ACK N OWLED G EM ENT
We appreciated for the clinical advice by professor Yan-Sheng Shan of National Cheng-Kung University. This article was edited by Wallace Academic Editing.

CO N FLI C T O F I NTE R E S T S
There is no conflict of interests in this work.

DATA AVA I L A B I L I T Y S TAT E M E N T
All the data in the work can be provided upon reasonable request.