Downregulation of GLYAT facilities tumor growth and metastasis and poor clinical outcomes through the PI3K/AKT/Snail pathway in human breast cancer

The Glycine N-acyltransferase (GLYAT) gene encodes a protein that catalyzes the transfer of acyl groups from acyl CoA to glycine, resulting in acyl glycine and coenzyme A. Aberrant GLYAT expression is associated with several malignant tumors, but its clinical importance in malignant tumors, especially human breast cancer (BC), has yet to be fully addressed. This study aims to evaluate the clinical function of GLYAT in BC patients.

Breast cancer (BC) amongst women has been on the rise and it is fast becoming the most common malignancy in this population [1]. Leading cancer statistics show that the incidence and mortality of breast cancer remain high worldwide [2,3]. Although surgery, chemoradiotherapy, endocrine therapy, and molecular targeted therapy have greatly improved the survival of BC patients, those with advanced BC are often left with very limited therapeutic options [4,5]. Many factors in uence the prognosis and probability of responding to systemic therapies (e.g., TNM stage, ER, PR, and HER2 status), but clinically, we have observed that even patients with the same TNM stage, molecular typing, and other diagnostic markers still experience varied prognoses despite being given the same treatment modalities. While early diagnosis and initiation of treatment are crucial in improving survival rates of this debilitating disease, there is an urgent need for a deeper understanding of its molecular biology in order to aid in the development of more personalized and targeted treatment.
The EMT involves the dedifferentiation of epithelial cells into mesenchymal cells [6,7]. Cells which have undergone EMT demonstrate more malignant features that promote metastasis, treatment resistance, as well as relapse [8][9][10]. The EMT participates in tumor invasion and metastasis through multiple pathways. Many studies have indicated the EMT is strongly linked to breast cancer development [11][12][13].
The EMT is regulated by multiple factors, among which the primary signaling pathway is the PI3K/AKT pathway [14].
The glycine N-acyltransferase (GLYAT) gene was rst discovered in the mitochondria of the bovine liver in 1953 and isolated in the human liver and kidneys in 1976 [15]. It contains more than 23,000 base pairs over six exons and is situated chromosome 11 at position 11q12 [16]. GLYAT catalyzes acyl group transfer from acyl CoA to glycine, resulting in acyl glycine and coenzyme A (acyl-CoA) [17]. Several catabolic and anabolic reactions are catalyzed by acyl-CoA esters [18]. Almost all catabolic reactions produce acyl-CoA, the product of which is an important source of oxidative phosphorylation and lipogenesis. Studies have shown that GLYAT expression is suppressed in human hepatocellular carcinomas and may be a critical molecule in the transition between the differentiation and carcinogenesis of liver cells [19]. Nevertheless, literature is scarce surrounding GLYAT expression and its impact on human breast cancer.
The current investigation uncovered markedly suppressed GLYAT expression in breast cancer cells and tissues, which correlated with poorer prognosis and highly malignant clinicopathologic features in individuals with breast cancer. Interestingly, the EMT pathway appeared to be augmented in the presence of GLYAT, working to decrease in vivo and in vitro breast cancer cell migration through alteration of the PI3K/ATK/Snail signaling pathway. In conclusion, our study is the rst of its kind to implicate GLYAT to be a breast cancer anti-oncogene. These ndings lay the foundation for future research on the biological behaviour and targeted therapy of BC.

Datasets and GLYAT expression analysis
The public data used in this study were obtained from UALCAN dataset (http://ualcan.path.uab.edu/index.html) [20], GEPIA (http://gepia.cancer-pku.cn/index.html) [21], and the Human Protein Atlas dataset (http://www.proteinatlas.org) [22]. For UALCAN data, we analyzed the heat map of GLYAT expression between normal and breast cancer samples; For GEPIA data, we scrutinized differently GLYAT expression level and survival data in breast cancer. The relationship between patient survival and GLYAT levels were also analyzed using information available from the Human Protein Atlas data.

Cell lines
Human BC cell lines MDA-MB-231, MCF-7, and SKRB-3 were supplied by The Stem Cell Bank, Chinese Academy of Sciences (Shanghai, China). Cells were maintained at 37°C in a humidi ed cubicle containing 5% CO 2 and 10% fetal bovine serum (FBS) in DMEM (all from Gibco, Carlsbad, CA, USA).

Plasmidformation of GLYAT knockdown and overexpression in breast cancer cells
The pGPU6/mCherry/Puro-shRNA-GLYAT plasmid was purchased from GenePharma Company (Shanghai, China). The pOGP-T2A-CKNeo-GLYAT overexpression plasmid was purchased from JTS Scienti c Company (Wuhan, China). The target sequences of GLYAT were indicated in (Additional le 1: Table S1). Lipofectamine 3000 (Thermo Fisher Scienti c, US) was used to transfect plasmids into cells as reported by the manufacturer's instructions. Cells transfected with GPU6/mCherry/Puro-shNC or pOGP-T2A-CKNeo-NC were used as controls. Transfected cells were screened using 3 ug/mL puromycin or 400 ug/mL G418 for 2 weeks. Stable cells were cultured in complete medium with 0.25 ug/mL puromycin or 100 ug/mL G418 (Beyotime, Nanjing, China). Positive clones were then selected and ampli ed for further analyses.

Cell proliferation assay
For colony forming experiments, six-well plates were used to house stable GLYAT-shRNA and NC cells.
The cells were maintained at a concentration of 500 cells/well for 12 days in DMEM supplemented with 10% FBS, with media changed once every three days. Cells were then methanol-xed and treated with crystal violet (Sigma, St Louis, MO, USA) prior to manual counting and photographing of the visible colonies. For soft agar experiments, Cells are harvested and pipetted well to become single-cell suspension in complete culture media in 1x 106 /ml. A mixture of 0.9 ml 4% soft-agar (Sigma) with 4.1 ml pre-warmed 10% FBS DMEM was added into a 60-mm culture dish to make the bottom layer. The top layer contained 3 x 104 cells in 3 ml of 10% FBS DMEM and 0.36% agar. The soft-agar colony dish was marked and placed at a 37 °C incubator for 3 weeks. cells were harvested and fully mixed to form a single-cell suspension in media of 1 x 10 6 /ml. The upper layer included 3 x 10 4 cells, 3 ml of 10% fetal bovine serum, DMEM, combined with 0.36% agar (Sigma), while the bottom layer contained 0.9 ml 4% soft-agar, 4.1 ml pre-heated 10% fetal bovine serum, and DMEM in a 60-mm culture dish. Labeled colony dishes were placed in an incubator at 37°C for 3 weeks.

Transwell migration assay
Costar chambers containing 8 μm pore inserts were used (Millipore-Sigma, Danvers, MA, USA). The top chamber was used to house suspended cells transfected with GLYAT KD, GLYAT OE, or NC in 200 μl of serum-free medium, while media with 20% FBS was applied to the bottom chamber. The upper chamber was removed after 36 hours of incubation, with the bottom cartridge xed with methanol and DAPIstained to determine the migrated number of cells using a microscope to visualize the cells across ve randomly selected elds (Olympus, Tokyo, Japan).
Immuno uorescence staining Cells were subjected into 8 well chamber slides (Millipore-Sigma, Danvers, MA, USA) and for 20 hours incubation then rinsed with PBS containing 10% FBS. Pre-cooled methanol was used to x the cells prior to further incubation with 0.2% Triton. BSA (5%) was used to block cells. The cells were next incubated overnight with primary antibodies E-cadherin, vimentin (1:200 dilution respectively) before being washed and re-incubated with uorescein-conjugated secondary antibodies (1:500 ZSGB-Bio, Beijing China) for one hour. Then, DAPI and a uorescence microscope (Olympus, Tokyo, Japan) were used for visualization and intensity analysis. For dewaxed sections, the process was similar like above, and primary antibodies dilution of E-cadherin, vimentin and p-AKT Ser473 were 1:100 respectively.

In vivo metastasis assay
The HFK BIOSCIENCE CO., LTD (Beijing, China) provided 5 weeks old BALB/c (nu/nu) female nude mice (18-20 g). These animals were divided into four cohorts with ve mice each. Food and drinking water were provided every day. All mice received subcutaneous injections into their left breast fat pads with 100 microliters of PBS containing 2.0 × 10 6 MCF7 cells with or without GLYAT OE and MDA-MB-231 cells with or without GLYAT KD. At time of MCF7 cells injection, E2 pellets (60-day release, 1.5 mg/pellet; Innovative Research of America,Sarasota, Florida USA) was implanted subcutaneously at the mammary fat pad.
Four days measurements and volume calculation ((mm 3 ) =length × width 2 /2) of the tumors were performed. The mice were sacri ced and underwent tumor removal after 24 days. The tumors were processed with 4% paraformaldehyde solution for immuno uorescence staining analysis.

Patients and samples
Two independent breast cancer patient cohorts from the Shengjing Hospital of CMU were collected for this investigation. The rst cohort consisted of 21 breast cancer patients (seven for Luminal A, seven for Luminal B, and seven for TNBC) and 20 normal controls. Samples from the cancer patients and normal controls were analyzed for GLYAT expression. The second cohort comprised of 310 chemoradiotherapynaïve breast cancer patients who had distinctive pathological diagnoses and possessed complete followup data from 2006 to 2008. Additional follow-up of these patients was carried out until September 2012.

Immunohistochemistry
The dewaxed sections were retrieved with Tris/EDTA (pH9.0) and quenched with 0.3% H 2 O 2 . Sections then underwent an overnight incubation with GLYAT primary antibody (1:100 Abcam, Cambridge, MA, USA). PBS was used to rinse the samples prior to repeat incubation with secondary biotinylated antibodies at room temperature for 45 minutes. Coloration was visualized with 3,3 '-diaminobenzidine tetrachoric acid (Sigma-Aldrich), Vector hematoxylin QS (Vector Laboratories) was reverse-stained, Zeiss Mirax MidiSlide scanner was analyzed, and image collection was performed with 3 CCD color cameras and Panoramic Audience (3DHISTECH, Budapest Hungary). H-score was calculated as the percentage of positive tumor cells multiplied by the intensity of staining (0, no staining; 1, weak; 2, mild to moderate; 3, strong staining) and the overall H-score ranged from 0 to 300. An H-score of more than 100 was set as cut-off value to de ne high/low level for GLYAT expression.

Statistical analysis
Data were analyzed using SPSS version 25.0 (SPSS, Inc., Chicago, IL, USA) and expressed as means ± SD. The Kaplan-Meier method was used to estimate survival curves. Cox regression model was applied to carry out univariate and multivariate statistical analysis. Comparison between two groups was assessed by Chi-square. P value of less than 0.05 was considered statistically signi cant.

Characterization of GLYAT expression pro ling in human BC
The heat map analysis of the UALCAN dataset showed lower expression of GLYAT protein and mRNA in BC tissues in contrast to healthy samples ( Fig. 1a and 1c). However, the expression of GLYAT was not statistically signi cant change in other cancer types, besides BRCA, LIHC and CHOL (Additional le 2: Figure S1), which indicated that GLYAT may have speci c function in breast cancer. Additionally, analysis of data published in another study in Nature [23] which compared mRNA expression levels of GLYAT in normal breast tissues with invasive ductal and invasive lobular carcinoma revealed that two breast carcinoma subtypes had lower GLYAT mRNA levels (Fig. 1b). A Kaplan-Meier analysis from GEPIA dataset and a scatter plot from the Human Protein Atlas dataset suggested that patients with low GLYAT expression in breast cancer tissues had worse overall survival (OS) (Fig. 1d-e), further highlighting the suppressive function of GLYAT in BC progression.

GLYAT suppresses BC cell proliferation and metastasis
Three typical breast cancer cell lines were scrutinized for GLYAT expression. These cell lines included SK-BR3 (Her-2 positive subtype), MDA-MB-231 (triple negative subtype) and MCF-7 (luminal A subtype). Western blotting showed that the MDA-MB-231 cell line had the highest GLYAT expression, while the MCF-7 cell line had the lowest amount (Fig. 2a). Therefore, these two cell lines above were selected for further experiments.
In vitro assays were carried out in MDA-MB-231 cells with GLYAT knockdown (KD) and overexpression (OE) in MCF-7 cells to determine the impact of GLYAT expression on breast cancer behavior (Fig. 2b-c). The tablet colony formation and soft agar colony formation experiments revealed that GLYAT KD in MDA-MB-231 cells signi cantly increased colony numbers in contrast to control cell lines ( Fig. 2d-e). Transwell migration assay revealed arti cially altered MDA-MB-231 cells with GLYAT KD also had stronger migratory abilities in contrast to the control cells (Fig. 2f). In line with these results, the converse was seen in MCF-7 cells with OE GLYAT (Fig. 2g). These results indicate that GLYAT may hinder the proliferative and metastatic abilities of breast cancer cells.

GLYAT suppresses the EMT phenotype in BC cells
Given the prominent role of EMT in tumor spread, we postulated that GLYAT may be involved in the EMT process. In vitro assays were performed in MDA-MB-231 with GLYAT KD and in MCF with GLYAT OE. Both vimentin and E-cadherin were analysed for their nucleic location via immuno uorescence assays. In GLYAT KD cells, we found an increase in vimentin but a decrease in E-cadherin expression ( Fig. 3a-b). Western blotting of EMT-related proteins was consistent with the immuno uorescence results, showing a reduction in E-cadherin, but increase in levels of vimentin, N-cadherin, and bronectin in MDA-MB-231 GLYAT KD cells (Fig. 3c-g). On the contrary, MCF-7 GLYAT OE cells demonstrated raised E-cadherin protein expression and lowered vimentin levels ( Fig. 3h-j). These results suggest that EMT formation could be suppressed by GLYAT.
GLYAT suppresses the EMT of BC cells via the PI3K/Akt/Snail pathway EMT process is highly regulated by the PI3k/Akt/Snail pathway. We therefore hypothesized that GLYAT might suppress breast cancer metastasis by modulation of this pathway. Western blot assays indicated that p-AKT, AKT, PI3K p58, and SNAI1 were signi cantly activated in cells with GLYAT KD (Fig. 4a-e). OE cells on the other hand, demonstrated markedly suppressed levels of p-AKT, AKT, PI3K p58, and SNAI1 in contrast to control cells (Fig. 4f-j). Our results indicate the involvement of the PI3K/Akt/Snail signaling pathway in GLYAT-mediated EMT suppression in BC cells.

GLYAT suppresses in vivoBC EMT, proliferation and metastasis
In vivo experiments were performed on nude mice that were received subcutaneous injection with either stable MDA-MB-231 GLYAT KD or MCF-7 GLYAT OE cells (Fig. 5a). The mice injected with stable GLYAT KD cells developed bigger and heavier tumors (Fig. 5b-d). In contrast, the mice injected with stable GLYAT OE MCF-7 cells had markedly smaller and lighter tumors (Fig. 5e-g). Then immuno uorescence assays for E-cadherin, vimentin, and p-AKT were performed using serial sections of mouse tumor tissues. These assays revealed that GLYAT silencing in MDA-MB-231 cells signi cantly decreased E-cadherin levels, but increased vimentin and p-AKT levels (Fig. 5h). In contrast, in GLYAT OE MCF-7 cells, E-cadherin was increased, whereas vimentin and p-AKT were decreased (Fig. 5i).
Lower GLYAT expression is correlated with poorer prognosis and malignant clinicopathologicalfeatures in human breast cancer tissues Immunohistochemical experiments found that breast cancer tissues possessed markedly decreased GLYAT expression in contrast to healthy breast tissue ( Fig. 6a-b). Additionally, GLYAT expression in BC tissues correlated positively with the degree of pathological differentiation. (Fig.6c). Based on these results, we conclude GLYAT may act as an anti-oncogene in BC.
We then studied the correlation between clinicopathological features and GLYAT expression in 310 patients who were recruited for this investigation. Table 1 demonstrates their characteristics. All individuals were women between 29 and 78 years old. No one had a previous history of malignancy and were chemoradiotherapy-naïve. The patients were grouped based on their GLYAT levels-low (n=143) or high (n=167) expression groups (tumor tissue/adjacent, normal samples). We found GLYAT expression correlated with TNM stage, histological grade, and Ki-67 status (Table 1; all P<0.05). Additionally, we did not uncover any relationship between GLYAT level and menopausal status, age, receptor status (HER, ER, PR), tumor size, node status, pathologic or molecular subtype (Table 1; all P>0.05). These ndings suggest GLYAT is associated with malignant clinicopathological characters of BC. GLYAT, Glycine N-acyltransferase; TNM, tumor-node-metastasis.
BC patients were analyzed for their GLYAT status and prognosis via Kaplan-Meier survival analysis as well as log-rank tests after being followed up for an average of 61.75 months (range, 9-77 months). We found that those with decreased GLYAT levels experienced poorer disease free survival (DFS) (Fig. 6d Table 2) and represented an independent risk factor of prognosis for BC. In the subgroup analyses based on Ki67, those with low GLYAT levels were found to have shorter DFS in comparison to those with high GLYAT, regardless of whether they had high or low Ki67. This difference was statistically signi cant in those with low Ki67 (P=0.014, Fig. 6i). Similarly, those with low GLYAT levels also experienced shorter DFS in contrast to those with high GLYAT, irrespective of ER+ or ER-status. There was a signi cantly statistical difference in those of ER+ status (P=0.011, Fig. 6j). The same pattern of ndings also applied to subgroup analyses based on PR. We also observed a statistically signi cant difference in the subgroup of PR+ patients (P=0.011, Fig. 6k). Likewise, HER2 status did not change the impact of low GLYAT on shorter DFS, with statistically signi cant changes observed in both HER+ or HERstatus groups (P=0.046 and P=0.045 , Fig. 6l).

Discussion
Advanced breast cancer is a condition that carries high morbidity and mortality, despite the existence of several diagnostic markers and treatment strategies [24]. Therefore, we sought to determine novel molecular components involved in the biology of BC that may be useful in the management of this condition.
A potential candidate is the GLYAT protein, which is involved in glycine conjugation of xenobiotics such as benzoic acid, and plays a role in many anabolic and catabolic reactions [25]. In cancerous cells, GLYAT functions as a key metabolite that inhibits glycine uptake or biosynthesis, impairing growth of these cells likely through inhibition of nucleic acid synthesis [26]. Previous reports have highlighted the role of this molecule in musculoskeletal growth and development as well as in hepatocellular carcinoma progression [19]. Nevertheless, there has been no study on its oncological effect in breast cancer.
Our series of experiments found that human breast cancer cells and tissues contained remarkably suppressed levels of GLYAT. We further uncovered that lower GLYAT level was associated with more malignant clinicopathological characteristics, including higher TNM stage, histological grade, and Ki-67 status. DFS of those with lower GLYAT levels were also shorter. Our ndings are consistent with information available in public databases. As breast cancer is related to hormone which has different molecular types, we then investigated the relation between GLYAT and patient prognosis based on their molecular subtypes, namely their HER, PR, ER, and Ki67 statuses. Lower GLYAT levels resulted in poorer DFS irrespective of molecular pro le, as demonstrated by statistically signi cant differences in those of the luminal A subgroup, the Ki67 low status subgroup, the ER + subgroup, the PR + subgroup, and both the HER + and HER-subgroups (all P < 0.05). We postulate that the lack of statistical signi cance for other molecular subtype subgroups, such as the Ki67 high status subgroup, the ER-subgroup, and the PRsubgroup, can be attributed to small sample sizes or an inadequate follow-up time. However, perhaps more importantly, ER and PR status are intricately linked to breast cancer initiation, development and prognosis. GLYAT may act in combination with ER or PR status to impact breast cancer prognosis.
Additional experiments are needed to clarify this question in further research.
We further discovered GLYAT to be an independent prognostic factor for BC. These ndings indicated that GLYAT acts as antioncogene and is associated with malignant clinicopathological features that may enhance breast cancer metastasis and progression. Our ndings based on DFS alone are suggestive of the potential of this molecule to prognosticate breast cancer. Our ndings are consistent with bioinformatics analyses and the data reported by Liu et al., who indicated that GLYAT was suppressed in human hepatocellular carcinomas [19]. Given that our study is the rst to analyze the correlation between GLYAT and breast cancer, future investigations and follow up investigations regarding OS information (not available in our study due to short follow-up time) are still required.
We further demonstrated that GLYAT regulates breast cancer migration and invasion via EMT modulation. EMT is reduced via alteration of the PI3K/ATK/Snail signaling pathway both in vitro and in vivo. The EMT process participate in cancer metastasis [27]. Numerous studies reported that EMT is regulated by several signaling pathways, for example Notch-, Wnt-, PI3K/Akt-and NF-κB-dependent pathways [28][29][30]. Previous studies have con rmed that the EMT enhances tumor cell mobility and invasiveness, thus, contributing to the development of chemotherapy resistance and metastasis [31,32]. Additionally, breast cancer recurrence and prognosis have been reported to be affected strongly by EMT [33]. We believe there may be other mechanisms in addition to the PI3K/AKT/Snail pathway by which GLYAT promotes metastasis that need to be further explored. One example by Ren  signaling pathway [34]. GLYAT downregulation suppresses JNK-dependent ROS activation, which is a key facilitator of several important biological functions.
Besides the impact on metastasis, the proliferation ability of BC cells was increased after inhibiting GLYAT both in vitro and in vivo, further highlighting the tumor suppressed role of GLYAT. Our research is the only study to illustrate the role of GLYAT in BC patients, revealing that down-regulated GLYAT induces EMT and tumor metastasis via PI3K/ATK/Snail signaling.

Conclusion
Collectively, our data revealed that GLYAT downregulated in human BC cells and tissues, and lower GLYAT expression was related to poor clinical outcomes. GLYAT suppresses BC cell proliferation and migration through EMT induction via the PI3K/ATK/Snail pathway in vitro and in vivo. Our study puts forth GLYAT as a novel biomarker in BC and may also represent a therapeutic target for BC treatment.   and KD2 compared with the scramble control. g Transwell assay revealed that the migratory ability was signi cantly decreased in MCF-7 cells GLYAT OE compared with scramble control. *p<0.05, **p<0.01.
Scale bars for E are 100μm and 400μm, and for F and G are 500μm. and KD2 compared with the scramble control. g Transwell assay revealed that the migratory ability was signi cantly decreased in MCF-7 cells GLYAT OE compared with scramble control. *p<0.05, **p<0.01.
Scale bars for E are 100μm and 400μm, and for F and G are 500μm.