Proteomic Analysis Reveals That Metformin Suppresses PSMD2, STIP1, and CAP1 for Preventing Gastric Cancer AGS Cell Proliferation and Migration

Abstract

Metformin is one of the most widely used anti-diabetic drugs in type-II diabetes treatment. The mechanism of decreasing blood glucose is believed to suppress hepatic gluconeogenesis by increasing muscular glucose uptake and insulin sensitivity. Recent studies suggest that metformin may reduce cancer risk; however, its anticancer mechanism in gastric cancers remains unclear. Here, we aim to evaluate the anticancer effects of metformin on human gastric adenocarcinoma (AGS) cells. Our results showed that metformin inhibited AGS cell proliferation in a dose-dependent manner. Using small-scale quantitative proteomics, we identified 177 differentially expressed proteins upon metformin treatment; among these, nine proteins such as 26S proteasome non-ATPase regulatory subunit 2 (PSMD2), stress-induced phosphoprotein 1 (STIP1), and adenylyl cyclase-associated protein 1 (CAP1) were significantly altered. We found that metformin induced cell cycle arrest at the G0/G1 phase, suppressed cell migration, and affected cytoskeleton distribution. Additionally, patients with highly expressed PSMD2, STIP1, and CAP1 have a poor clinical outcome. Our study provides a novel view of developing therapies for gastric cancer.

Introduction

Gastric cancer is the third common cancer in the world, accounting for 8% of total cancer and 10% of total deaths.¹ Gastric cancer is asymptomatic in early stages. Most of the patients are diagnosed with gastric cancer at the terminal stage of cancer. Therefore, it is one of the reasons why the 5 yr survival rate of gastric cancer patients is less than 30%.² Surgical resection is still the major curative treatment for gastric cancer; however, the survival rate remains low due to rapid metastasis and high recurrence rates. The recurrence rate of gastric cancer is 40–70%,³ and it is diagnosed within 20–28 months.⁴ Additional perioperative chemotherapy or adjuvant chemoradiation have shown a poor survival rate in gastric cancer;⁵ therefore, new therapies are urgently required.

Metformin is a biguanide derivative that originates from goat’s rue. Now, it serves as the first line and widely used oral anti-diabetic drug in the treatment of type-II diabetes. A previous study suggests that metformin can reduce metabolic risk factors and sensitize insulin receptors, especially in overweight youths with type-I diabetes.^6,7 Metformin decreases the blood glucose level through increasing muscular glucose uptake, reducing glucose production in the liver, and enhancing insulin receptor sensitivity.⁸⁻¹¹ Thus, it regulates hepatic gluconeogenesis and glycogenolysis. In recent years, the relationship between metformin and cancer has been discussed in several studies. Evans et al. found that metformin may reduce the risk of cancer in a case-control study.¹² Several epidemiological studies also suggest that diabetic patients who use metformin have a lower cancer incidence than those who use other anti-diabetic treatments.^7,13 Recent studies show that metformin inhibits cancer cell proliferation in the breast,¹⁴ lung,¹⁵ and colon¹⁶ cancer cell lines. Some studies indicate that metformin mediates the activation of 5’ adenosine monophosphate-activated protein kinase (AMPK) and results in the inhibition of breast cancer cell and intestinal polyp growth with decreased mTOR and S6K activity, which play an essential role in protein synthesis.^17,18 Metformin has the ability to inhibit cancer viability through the inhibition of mTOR in an AMPK-independent manner.¹⁹ Therefore, metformin treatment in cancer research has become a suitable and interesting topic due to its safety, low cost, and the property of modulating energy metabolism.²⁰

Most of the biological functions, regulatory switches, and signal transfections are controlled by multiple proteins rather than a single one.²¹ Proteomics can provide an integrative view of protein characterization and quantification in organisms under defined conditions.²² Therefore, proteomics is applied in many fields such as drug discovery,²³⁻²⁵ cancer therapy,^26,27 and uncovering chemical-induced biological mechanisms.²⁸⁻³⁰ One of the proteomic techniques is called isobaric tagging for relative and absolute quantitation (iTRAQ), which was developed in 2004.³¹ iTRAQ is a technique that applied a multiplexed isobaric chemical tag, which allows labeling four groups of protein samples at a time. Relative quantification is achieved by calculating the peak areas for either four MS/MS reporter ions (ion range from 114 to 117 Da).³² In this study, we applied iTRAQ to identify and quantify metformin-regulated proteins. Furthermore, we performed several functional assays to show that metformin-regulated proteins were involved in those biological functions (Figure 1).

Figure 1.

Experimental design of this study. AGS cells were treated with metformin, proteins were labeled with isobaric tags for absolute quantitation, analyzed by mass spectrometry, and then validated.

Results and Discussion

Metformin Inhibits the Proliferation of Human Gastric Cancer AGS Cells

To determine whether metformin affects human gastric cancer cell growth, we analyzed the effect of metformin on cell proliferation using a human gastric cancer cell line, AGS. Cells were treated with different concentrations of metformin for 24, 48, and 72 h. As shown in Figure 2A, metformin inhibits the cell proliferation in a dose-dependent manner. We also observed similar results in the colony formation assay (Figure 2B). To further examine whether metformin inhibition of cell proliferation is reversible, we changed the medium after metformin treatment for 24 h (Figure 2C). We observed that cell proliferation was still suppressed when metformin had been removed (Figure 2D). The results suggest that metformin might irreversibly inhibit AGS cell proliferation. Together, metformin inhibits the AGS cell proliferation, and the effect might be irreversible.

Figure 2.

Metformin dose-dependently inhibits AGS cell proliferation and anchorage-dependent growth. (A) MTT assay results in which AGS cells were treated with different concentrations of metformin. (B) Colony formation assay results in which AGS cells were treated with 10 mM metformin or control for 7 days. The number of colonies was counted in triplicate experiments. Data are presented as mean ± S.D. (n = 3). * and ** indicate that p value <0.05 and <0.001, respectively. (C) Experimental design to check whether metformin-induced effect is reversible. Cells were seeded for 24 h and then the medium was changed with complete medium and medium containing metformin. For the control group and one of the Met-pretreated groups, the medium was changed with fresh complete medium after 48 h. For the Met-pretreated + Met group, the medium was changed with a medium that contained metformin after 48 h. (D) After 48 h of metformin treatment, AGS cells were incubated with or without metformin and analyzed for the cell proliferation using RTCA with an E-plate for another 72 h. Control: cells without any metformin treatment; Met-pretreated cells: cells pretreated with metformin for 48 h and then removed from metformin exposure; Met-pretreated cells + Met: cells pre-treated with metformin for 48 h and incubated in metformin. The x-axis indicates the normalized cell index and y-axis indicates the time in hour. Data are presented as mean ± S.D. (n = 3).

iTRAQ-Based Proteomics Reveals Metformin-Regulated Proteins

To investigate the differential proteins induced by metformin, we performed the iTRAQ-based quantitative proteomic approach, and the workflow is illustrated in Figure 3A. Cells were treated with metformin or control for 72 h. Proteins extracted from metformin-treated and control AGS cells were digested to peptides and labeled with iTRAQ. Duplicates of control and treatment showed a high correlation in iTRAQ analysis. The correlation coefficient R² was 0.9206 between the control duplicates and 0.9461 between the treatment duplicates (Figure 3B,C).

Figure 3.

iTRAQ-based quantitative proteome in control and metformin-treated AGS cells. (A) Schematic representation of the experimental strategy for quantitative global proteomic profiling in response to metformin in AGS cells. Protein extracts obtained from the transfected cells were digested, iTRAQ-labeled, SCX-fractionated, and analyzed by mass spectrometry. (B,C) iTRAQ analysis showing high reproducibility in the duplicate sample. The correlation of iTRAQ intensity between duplicate samples (control: iTRAQ114 and iTRAQ115; metformin: iTRAQ116 and iTRAQ117). (D) Distribution of protein ratio in the control and metformin-treated samples. Protein ratios of total 177 proteins were log₂-transformed and compared to the cutoff value to identify differentially expressed proteins. There were five downregulated proteins and four upregulated proteins in metformin treatment group compared to control. (E) Protein expression level of cancer-associated proteins identified by iTRAQ. Cells were treated with 10 mM metformin for 48 h, then harvested, and proteins extracted. The protein levels of PSMD2, STIP1, and CAP1 identified by iTRAQ were verified by western blotting. (F) Overall survival of the low expression (blue) and high expression (res) of (A) PSMD2, (B) STIP1, and (C) CAP1. Patients (n = 192) were classified according to low expression and high expression of specific genes. Data were obtained from Gene Expression Omnibus accession number GSE15459, and the P-values were obtained from the log-rank (Mantel–Haenszel) test.

A total of 177 proteins were identified. Nine differential proteins included five downregulated proteins and four upregulated proteins between the control and treatment (Figure 3D and Table 1). Five downregulated proteins include t-complex protein 1 subunit epsilon (CCT5), 26S proteasome non-ATPase regulatory subunit 2 (PSMD2), stress-induced phosphoprotein 1 (STIP1), polypyrimidine tract-binding protein 1 (PTBP1), and adenylyl cyclase-associated protein 1 (CAP1). The western blotting showed that PSMD2 (0.76-fold), STIP1 (0.22-fold) and CAP1 (0.61-fold) were downregulated in the protein sample of metformin treatment cells, which confirmed the results in iTRAQ analysis (Figure 3E).

Table 1. Total Protein Identified in Control and Metformin-Treated AGS Cells.

accession no	gene symbol	protein name	mass (Da)	protein score	no. of peptidesa	Log2 (protein ratio) ± S.D. (treatment/control)
H15_HUMAN	HIST1H1B	histone H1.5	22566	77	2	–1.05 ± 0.35
CNTRL_HUMAN	CNTRL	centriolin	268720	42	1	–1.03 ± 0.28
TCPH_HUMAN	CCT7	T-complex protein 1 subunit eta	59329	81	1	–0.98 ± 0.29
TIF1B_HUMAN	TRIM28	transcription intermediary factor 1-beta	88493	73	3	–0.96 ± 0.36
LA_HUMAN	SSB	Lupus La protein	46808	125	4	–0.95 ± 0.89
LMNB1_HUMAN	LMNB1	Lamin-B1	66368	43	1	–0.89 ± 0.36
U2AF2_HUMAN	U2AF2	splicing factor U2AF 65 kDa subunit	53467	52	2	–0.88 ± 0.61
MDHM_HUMAN	MDH2	malate dehydrogenase, mitochondrial	35481	162	3	–0.86 ± 0.48
2AAA_HUMAN	PPP2R1A	serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A alpha isoform	65267	156	4	–0.83 ± 1.27
SERPH_HUMAN	SERPINH1	serpin H1	46411	153	3	–0.71 ± 0.22
G6PI_HUMAN	GPI	glucose-6-phosphate isomerase	63107	310	7	–0.68 ± 0.7
MBB1A_HUMAN	MYBBP1A	Myb-binding protein 1A	148762	70	2	–0.64 ± 0.43
RPN2_HUMAN	RPN2	dolichyl-diphosphooligosaccharide--protein glycosyltransferase 2	69241	81	2	–0.63 ± 1.03
1433E_HUMAN	YWHAE	14-3-3 protein epsilon	29155	327	4	0.36 ± 0.21
RL27A_HUMAN	RPL27A	60S ribosomal protein L27a	16551	39	1	0.37 ± 0.42
IPYR_HUMAN	PPA1	inorganic pyrophosphatase	32639	47	1	–0.57 ± 0.07
RL9_HUMAN	RPL9	60S ribosomal protein L9	21850	39	1	–0.57 ± 1.17
DX39A_HUMAN	DDX39A	ATP-dependent RNA helicase DDX39A	49098	42	1	–0.52 ± 0.4
HNRPF_HUMAN	HNRNPF	heterogeneous nuclear ribonucleoprotein F	45643	144	4	–0.51 ± 0.11
H2A1B_HUMAN	HIST1H2AB	histone H2A type-1-B/E	14127	185	2	–0.5 ± 0.22
RS16_HUMAN	RPS16	40S ribosomal protein S16	16435	119	2	–0.49 ± 0.19
KBTB3_HUMAN	KBTBD3	Kelch repeat and BTB domain-containing protein 3	69350	42	1	–0.48 ± 0.14
PGK1_HUMAN	PGK1	phosphoglycerate kinase 1	44586	106	3	–0.46 ± 0.31
RS28_HUMAN	RPS28	40S ribosomal protein S28	7836	42	1	–0.45 ± 0.25
6PGD_HUMAN	PGD	6-phosphogluconate dehydrogenase, decarboxylating	53106	158	5	–0.44 ± 0.31
K2C1_HUMAN	KRT1	keratin, type-II cytoskeletal 1	65999	602	16	–0.42 ± 0.17
IMB1_HUMAN	KPNB1	importin subunit beta-1	97108	237	5	–0.41 ± 0.65
FUS_HUMAN	FUS	RNA-binding protein FUS	53394	53	1	–0.39 ± 0.45
IF4A1_HUMAN	EIF4A1	eukaryotic initiation factor 4A-I	46125	204	9	–0.39 ± 0.17
PCBP2_HUMAN	PCBP2	poly(rC)-binding protein 2	38556	99	3	–0.38 ± 0.77
S10AB_HUMAN	S100A11	protein S100-A11	11733	136	2	–0.36 ± 0.54
1433B_HUMAN	YWHAB	14-3-3 protein beta/alpha	28065	377	3	–0.36 ± 0.15
ROA2_HUMAN	HNRNPA2B1	heterogeneous nuclear ribonucleoproteins A2/B1	37407	431	8	–0.22 ± 0.09
H12_HUMAN	HIST1H1C	histone H1.2	21352	125	3	–0.21 ± 0.36
PRDX6_HUMAN	PRDX6	peroxiredoxin-6	25019	143	4	–0.21 ± 0.11
GRP75_HUMAN	HSPA9	stress-70 protein, mitochondrial	73635	209	4	–0.21 ± 0.24
TBB4B_HUMAN	TUBB4B	tubulin beta-4B chain	49799	396	2	0.44 ± 0.21
CLIC1_HUMAN	CLIC1	chloride intracellular channel protein 1	26906	167	3	0.44 ± 0.12
RS25_HUMAN	RPS25	40S ribosomal protein S25	13734	68	3	0.44 ± 0.25
RL4_HUMAN	RPL4	60S ribosomal protein L4	47667	102	3	0.46 ± 0.16
TBB5_HUMAN	TUBB	tubulin beta chain	49639	688	7	–0.34 ± 0.3
RS9_HUMAN	RPS9	40S ribosomal protein S9	22578	77	2	0.54 ± 0.29
IF5A1_HUMAN	EIF5A	eukaryotic translation initiation factor 5A-1	16821	49	1	0.54 ± 0.55
DDX3X_HUMAN	DDX3X	ATP-dependent RNA helicase DDX3X	73198	209	5	0.48 ± 0.41
RS10_HUMAN	RPS10	40S ribosomal protein S10	18886	102	2	–0.1 ± 0.11
HNRPU_HUMAN	HNRNPU	heterogeneous nuclear ribonucleoprotein U	90528	54	2	–0.09 ± 0.23
TCPQ_HUMAN	CCT8	T-complex protein 1 subunit theta	59583	109	3	0.55 ± 0.35
XRCC5_HUMAN	XRCC5	X-ray repair cross-complementing protein 5	82652	80	1	0.62 ± 0.65
SMD3_HUMAN	SNRPD3	small nuclear ribonucleoprotein Sm D3	13907	68	2	0.9 ± 1.06
GRP78_HUMAN	HSPA5	78 kDa glucose-regulated protein	72288	392	8	–0.21 ± 0.11
4F2_HUMAN	SLC3A2	4F2 cell–surface antigen heavy chain	67952	88	2	–0.21 ± 0.8
TALDO_HUMAN	TALDO1	transaldolase	37516	65	1	–0.2 ± 1.13
EF2_HUMAN	EEF2	elongation factor 2	95277	555	16	–0.19 ± 0.16
HMGB1_HUMAN	HMGB1	high mobility group protein B1	24878	40	1	–0.19 ± 0.04
RLA0L_HUMAN	RPLP0P6	60S acidic ribosomal protein P0-like	34343	53	1	–0.16 ± 0.51
HSPB1_HUMAN	HSPB1	heat shock protein beta-1	22768	90	3	–0.13 ± 0.11
TCPZ_HUMAN	CCT6A	T-complex protein 1 subunit zeta	57988	221	6	–0.13 ± 0.26
HSP7C_HUMAN	HSPA8	heat shock cognate 71 kDa protein	70854	748	14	–0.13 ± 0.05
SET_HUMAN	SET	protein SET	33469	72	1	–0.12 ± 0.42
KPYM_HUMAN	PKM	pyruvate kinase isozymes M1/M2	57900	1101	31	–0.12 ± 0.11
RSSA_HUMAN	RPSA	40S ribosomal protein SA	32833	57	1	–0.12 ± 0.27
VDAC1_HUMAN	VDAC1	voltage-dependent anion-selective channel protein 1	30754	142	2	–0.11 ± 0.61
RS27A_HUMAN	RPS27A	Ubiquitin-40S ribosomal protein S27a	17953	160	6	–0.11 ± 0.08
CALR_HUMAN	CALR	calreticulin	48112	52	1	–0.1 ± 0.49
H31T_HUMAN	HIST3H3	histone H3.1t	15499	80	6	–0.1 ± 0.07
TKT_HUMAN	TKT	transketolase	67835	239	4	–0.1 ± 0.46
RS3_HUMAN	RPS3	40S ribosomal protein S3	26671	60	3	–0.09 ± 0.08
PAIRB_HUMAN	SERBP1	plasminogen activator inhibitor 1 RNA-binding protein	44938	74	2	–0.08 ± 0.96
HNRDL_HUMAN	HNRPDL	heterogeneous nuclear ribonucleoprotein D-like	46409	76	2	–0.08 ± 0.18
GBLP_HUMAN	GNB2L1	guanine nucleotide-binding protein subunit beta-2-like 1	35055	66	2	–0.08 ± 0.26
H4_HUMAN	HIST1H4A	histone H4	11360	367	12	–0.08 ± 0.03
RS26_HUMAN	RPS26	40S ribosomal protein S26	13007	55	1	–0.08 ± 0.36
RS13_HUMAN	RPS13	40S ribosomal protein S13	17212	127	3	–0.07 ± 0.13
K1C19_HUMAN	KRT19	keratin,type-I cytoskeletal 19	44079	264	6	–0.07 ± 0.08
TBA1B_HUMAN	TUBA1B	tubulin alpha-1B chain	50120	1224	26	–0.07 ± 0.05
HNRPK_HUMAN	HNRNPK	heterogeneous nuclear ribonucleoprotein K	50944	219	7	–0.06 ± 0.11
RS3A_HUMAN	RPS3A	40S ribosomal protein S3a	29926	76	2	–0.05 ± 0.26
DDX5_HUMAN	DDX5	probable ATP-dependent RNA helicase DDX5	69105	85	3	–0.04 ± 0.16
H2B1D_HUMAN	HIST1H2BD	histone H2B type-1-D	13928	617	15	–0.04 ± 0.04
SAHH_HUMAN	AHCY	adenosylhomocysteinase	47685	101	4	–0.03 ± 0.19
LDHA_HUMAN	LDHA	l-lactate dehydrogenase A chain	36665	532	14	–0.02 ± 0.21
PDIA3_HUMAN	PDIA3	protein disulfide-isomerase A3	56747	247	4	–0.02 ± 0.23
HS90B_HUMAN	HSP90AB1	heat shock protein HSP 90-beta	83212	957	10	–0.01 ± 0.18
G3P_HUMAN	GAPDH	glyceraldehyde-3-phosphate dehydrogenase	36030	1036	24	0 ± 0.02
ENOA_HUMAN	ENO1	alpha-enolase	47139	711	20	0.01 ± 0.04
EF1G_HUMAN	EEF1G	elongation factor 1-gamma	50087	280	10	0.02 ± 0.14
CALX_HUMAN	CANX	calnexin	67526	70	2	0.02 ± 0.23
LDHB_HUMAN	LDHB	l-lactate dehydrogenase B chain	36615	293	10	0.02 ± 0.05
PPIA_HUMAN	PPIA	peptidyl-prolyl cis–trans isomerase A	18001	269	5	0.03 ± 0.1
EF1A1_HUMAN	EEF1A1	elongation factor 1-alpha 1	50109	328	14	0.05 ± 0.09
CH60_HUMAN	HSPD1	60 kDa heat shock protein, mitochondrial	61016	1137	26	0.05 ± 0.09
ROA3_HUMAN	HNRNPA3	heterogeneous nuclear ribonucleoprotein A3	39571	105	2	0.06 ± 0.13
LMNA_HUMAN	LMNA	prelamin-A/C	74095	147	8	0.06 ± 0.05
ACTG_HUMAN	ACTG1	actin, cytoplasmic 2	41766	1188	31	0.07 ± 0.02
TERA_HUMAN	VCP	transitional endoplasmic reticulum ATPase	89266	220	4	0.08 ± 0.52
HNRPM_HUMAN	HNRNPM	heterogeneous nuclear ribonucleoprotein M	77464	66	2	0.08 ± 1.67
ACTN4_HUMAN	ACTN4	alpha-actinin-4	104788	397	2	0.08 ± 0.28
GDIB_HUMAN	GDI2	Rab GDP dissociation inhibitor beta	50631	263	4	0.08 ± 0.68
K1C10_HUMAN	KRT10	keratin, type-I cytoskeletal 10	58792	741	13	0.1 ± 0.1
RAB10_HUMAN	RAB10	Ras-related protein Rab-10	22527	108	2	0.12 ± 0.85
NPM_HUMAN	NPM1	nucleophosmin	32555	427	16	0.13 ± 0.07
EZRI_HUMAN	EZR	ezrin	69370	184	9	0.13 ± 0.16
PSB3_HUMAN	PSMB3	proteasome subunit beta type-3	22933	198	4	0.13 ± 0.08
XPO2_HUMAN	CSE1L	exportin-2	110346	169	5	0.13 ± 0.1
ANXA1_HUMAN	ANXA1	annexin A1	38690	44	1	0.13 ± 0.37
PUR6_HUMAN	PAICS	multifunctional protein ADE2	47049	45	2	0.14 ± 0.18
RAB1A_HUMAN	RAB1A	ras-related protein Rab-1A	22663	72	1	0.14 ± 1.3
RSMB_HUMAN	SNRPB	small nuclear ribonucleoprotein-associated proteins B and B′	24594	39	1	0.15 ± 0.13
RS5_HUMAN	RPS5	40S ribosomal protein S5	22862	134	2	0.16 ± 0.7
RS23_HUMAN	RPS23	40S ribosomal protein S23	15798	75	2	0.16 ± 0.18
TAGL2_HUMAN	TAGLN2	transgelin-2	22377	202	5	0.17 ± 0.15
NP1L1_HUMAN	NAP1L1	nucleosome assembly protein 1-like 1	45346	262	5	0.17 ± 0.49
NUCL_HUMAN	NCL	nucleolin	76568	428	14	0.18 ± 0.18
K1C18_HUMAN	KRT18	keratin, type-I cytoskeletal 18	48029	1018	22	0.19 ± 0.01
DYHC1_HUMAN	DYNC1H1	cytoplasmic dynein 1 heavy chain 1	532072	48	1	0.21 ± 0.94
PROF1_HUMAN	PFN1	profilin-1	15045	164	4	0.22 ± 0.27
COF1_HUMAN	CFL1	cofilin-1	18491	269	5	0.22 ± 0.18
FLNA_HUMAN	FLNA	filamin-A	280564	197	5	0.22 ± 0.21
RS11_HUMAN	RPS11	40S ribosomal protein S11	18419	88	2	0.22 ± 0.12
ANXA2_HUMAN	ANXA2	annexin A2	38580	820	24	0.23 ± 0.15
1433Z_HUMAN	YWHAZ	14-3-3 protein zeta/delta	27728	473	6	0.24 ± 0.38
TCPG_HUMAN	CCT3	T-complex protein 1 subunit gamma	60495	95	3	0.26 ± 0.02
FAS_HUMAN	FASN	fatty acid synthase	273254	369	9	0.26 ± 0.6
K1C9_HUMAN	KRT9	keratin, type-I cytoskeletal 9	62027	86	1	0.28 ± 0.27
RL10A_HUMAN	RPL10A	60S ribosomal protein L10a	24816	71	2	0.29 ± 0.04
TPIS_HUMAN	TPI1	triosephosphate isomerase	30772	80	3	0.31 ± 0.25
PHB_HUMAN	PHB	prohibitin	29786	46	1	0.32 ± 1.46
PDIA1_HUMAN	P4HB	protein disulfide-isomerase	57081	195	7	0.34 ± 0.21
K2C8_HUMAN	KRT8	keratin, type-II cytoskeletal 8	53671	1343	33	0.34 ± 0.01
IQGA1_HUMAN	IQGAP1	ras GTPase-activating-like protein IQGAP1	189134	97	2	0.35 ± 0.65
RS8_HUMAN	RPS8	40S ribosomal protein S8	24190	66	2	0.35 ± 0.46
GTR1_HUMAN	SLC2A1	solute carrier family 2, facilitated glucose transporter member 1	54049	59	2	0.38 ± 0.16
RL6_HUMAN	RPL6	60S ribosomal protein L6	32708	142	3	0.38 ± 0.72
RS4X_HUMAN	RPS4X	40S ribosomal protein S4, X isoform	29579	134	4	0.38 ± 0.22
RL13A_HUMAN	RPL13A	60S ribosomal protein L13a	23562	54	1	0.43 ± 0.2
PGAM1_HUMAN	PGAM1	phosphoglycerate mutase 1	28786	133	4	–0.6 ± 0.5
MYH9_HUMAN	MYH9	myosin-9	226392	176	4	–0.59 ± 0.53
RL15_HUMAN	RPL15	60S ribosomal protein L15	24131	51	1	–0.59 ± 0.85
TRY1_HUMAN	PRSS1	trypsin-1	26541	54	1	–0.58 ± 1.18
DHX9_HUMAN	DHX9	ATP-dependent RNA helicase A	140869	447	12	–0.58 ± 1.34
NDKA_HUMAN	NME1	nucleoside diphosphate kinase A	17138	117	4	–0.02 ± 0.29
HS90A_HUMAN	HSP90AA1	heat shock protein HSP 90-alpha	84607	973	16	–0.02 ± 0.07
RL17_HUMAN	RPL17	60S ribosomal protein L17	21383	60	1	0.79 ± 0.21
LPPRC_HUMAN	LRPPRC	leucine-rich PPR motif-containing protein, mitochondrial	157805	141	4	0.82 ± 0.52
RL1D1_HUMAN	RSL1D1	ribosomal L1 domain-containing protein 1	54939	59	1	0.82 ± 0.79
ENPL_HUMAN	HSP90B1	endoplasmin	92411	146	3	0.08 ± 0.1
PRDX1_HUMAN	PRDX1	peroxiredoxin-1	22096	188	8	0.09 ± 0.21
ADT2_HUMAN	SLC25A5	ADP/ATP translocase 2	32831	79	4	0.09 ± 0.12
PP1A_HUMAN	PPP1CA	serine/threonine-protein phosphatase PP1-alpha catalytic subunit	37488	148	2	0.19 ± 1.04
MOT1_HUMAN	SLC16A1	monocarboxylate transporter 1	53909	42	1	0.2 ± 0.32
IPO7_HUMAN	IPO7	importin-7	119440	80	3	0.62 ± 0.43
HSP71_HUMAN	HSPA1A	heat shock 70 kDa protein 1A/1B	70009	207	1	0.64 ± 0.75
K22E_HUMAN	KRT2	keratin, type-II cytoskeletal 2 epidermal	65393	103	1	0.65 ± 0.35
RS20_HUMAN	RPS20	40S ribosomal protein S20	13364	52	1	0.74 ± 0.79
PSMD6_HUMAN	PSMD6	26S proteasome non-ATPase regulatory subunit 6	45502	65	2	0.79 ± 0.95

^aThe number of nondegenerate (unique) peptides used for quantitation in two technical replicates.

To understand whether the expressions of PSMD2, STIP1, and CAP1 are associated with patient survival, the Kaplan–Meier survival analysis was performed. We classified the patients into low expression and high expression. We observed that patients with highly expressed PSMD2, STIP1, and CAP1 had a poorer survival rate with statistical significance (Figure 3F).

Human CAP1 plays a crucial role in the actin turnover.³³ CAP1 knockdown shows an abnormal cell morphology and reduces cell migration in mammalian cells.³⁴ These studies and our results suggest that metformin induces inhibition of cell migration and abnormal cell morphology by reducing the CAP1 level. PSMD2 is a subunit of the 19S regulator of 26S proteasome, which regulates the proteasomal activity and cell growth by arresting the G1 phase of cell cycle in the lung cancer cell line. It is a signature of poor prognosis and metastasis phenotype in lung cancer patients.^35,36 siPSMD2 results in cell cycle arrest in breast and liver cancers and the inhibition of tumorigenesis.^37,38 STIP1 is an oncoprotein that stimulates ovarian cancer cell proliferation by activating the SMAD-ID3 signaling pathways.³⁹ A recent study suggests that siSTIP1 disrupts cellular migration and cell growth in lung cancer cells.⁴⁰ These results suggest that metformin might reduce protein levels of STIP1 and PSMD2 to inhibit AGS cell proliferation.

Metformin Blocks the Cell Cycle in the G0/G1 Phase and Inhibits Migration

According to our iTRAQ-based proteome, several differential proteins, including PSMD2, CAP1, and STIP1, are responsible for cell proliferation and cell cycle. We have found that metformin decreased cell viability previously (Figure 2A). To determine the underlying mechanism of cell growth inhibition, metformin-treated AGS cells were analyzed with flow cytometry. Metformin-treated cells showed an increasing G0/G1 phase and failed to enter the S phase (Figure 4A). To determine whether metformin affects cell cycle regulatory proteins, we examined the expression level of the cell cycle regulatory proteins of AGS cells after the treatment of metformin. CDK4 and CDK6 are the members of the cyclin-dependent protein kinase (CDK) family, which are essential for cell cycle G1 phase progression and G1–S phase transition.^41,42 CDK4 and CDK6 proteins’ expression level was significantly reduced in metformin-treated AGS cells (Figure 4B). These data suggested that metformin arrested the cell cycle in the G0/G1 phase by reducing CDK4 and CDK6. Many studies had been reported that the inhibition of cell cycle G1 regulatory protein inhibited cell proliferation in various cancers.^43,44 Therefore, these results suggest that metformin inhibits AGS cell proliferation and block cell cycle at the G1 phase through decreasing the protein levels of CDK4 and CDK6.

Figure 4.

Metformin blocks the AGS cell cycle at the G0/G1 phase and affects the expression level of cell cycle regulatory proteins. (A) Cells were starved for 24 h, then treated with 10 mM metformin for 48 h and harvested for flow cytometry analysis or protein extraction. The percentage of each cell cycle phase was analyzed by FCS Express. (B) Expression of cell cycle-related proteins CDK4 and CDK6. (C) After 48 h of metformin treatment, AGS cells were analyzed for cell migration using the transwell assay for another 9 h. (D) Quantification of the transwell assay. Data are presented as mean ± S.D. (n = 3). ** indicates that the p value <0.001. (E) After 48 h of metformin treatment, AGS cells were incubated with or without metformin treatment and analyzed for the cell migration using RTCA with a CIM plate for another 9 h. Data are presented as mean ± S.D. (n = 3) (F) Metformin induces reorganization of cytoskeletons in AGS cells. Cells were seeded on the coverslips in the six-well plate, incubated for 24 h, and then treated with 10 mM metformin for 48 h. After treatment, cells were fixed, stained, and visualized with a fluorescence microscope.

To compare our results with previous studies, we have surveyed whether the proteomic results of metformin-treated gastric cancer have been published. We only found the secretome research of metformin-treated gastric tumor-associated fibroblasts (TAFs). Chen et al. suggested that TAFs are essential for the gastric cancer microenvironment.⁴⁵ Factors involved in the functional enrichment of differential proteins of the secretome were cytoskeleton organization and cell cycle arrest. A previous study showed the enrichment of metformin-regulated proteins in cell cycle arrest and antiproliferation of breast cancer cell lines using deep proteomic techniques.¹⁷ The publications mentioned above found a common biological function that metformin induced cell cycle arrest and antiproliferation, resulting in cell death.

To evaluate whether metformin induces human gastric cancer cell apoptosis, we applied flow cytometry analysis. AGS cells were incubated with or without metformin for 72 h. We found that the percentages of the late apoptotic cells showed no significant difference between the control (8.84%) and metformin (9.16%)-treated cells (Figure S1A). Similarly, cells stained with 4′,6-diamidino-2-phenylindole (DAPI) did not show the difference in DNA condensation, one of the most important criteria to identify apoptotic cells,⁴⁶ between the control and metformin-treated cells (Figure S1B). These results showed that metformin did not induce apoptosis in AGS cells.

The migration and invasion of cancer cells is an initial step in tumor metastasis which is the most frequent cause of death for patients with cancer,⁴⁷ Therefore, we would like to determine if metformin affects cell migration ability in human gastric cancer cells. In the transwell assay, the cells that migrated across to the transwell filter were stained and counted. There was a remarkably reduced number of migratory cells in metformin treatment group compared with control (Figure 4C,D). Similarly, metformin inhibited the migration of AGS cells as assessed by real-time cell analysis (RTCA) (Figure 4E). These results indicate that metformin inhibits the migration ability of AGS cells.

Metformin Affects the Reorganization of the Actin Filament and the Distribution of Vinculin

Cell migration requires various cell-shape changes, which implicate the reorganization of the actin filament and distribution of focal adhesion proteins, such as vinculin.⁴⁸ The interaction between vinculin and membrane-bound β-catenin directly affects colorectal cancer metastasis.⁴⁹ Here, we examined the formation and distribution of the actin filament and vinculin in AGS cells. In control cells, the actin filament formed stress fiber formation, and the vinculin co-localized with F-actin at a distal end of the stress fiber. In contrast, metformin abolished the stress fiber formation of actin, and most of the vinculins were distributed around the cells (Figure 4F). Besides, the cell shape became more spindle-like and bigger in metformin treatment than in control. These data suggested that metformin affects the reorganization of the actin filament and distribution of vinculin, resulting in cell migration inhibition. Similar effects were found in hepatocellular carcinoma and cervical cancer cells through regulating the AMPK signaling pathway.^50,51 AMPK is known for the upstream regulation of vinculin and other cytoskeletons.⁵² Previous research has shown that metformin-induced differential proteins were enriched in the mTOR signaling pathway, which regulated migration.⁵³ We have found that metformin abolished actin stress fiber formation and changed the distribution of vinculin in AGS cells. These results suggest that metformin can inhibit cancer cell migration and enlarge the cell shape by affecting the distribution of focal adhesion and arrangement of actin.

Conclusions

In this study, we proposed the biological effects that are induced by metformin in Figure S2. Metformin-treated AGS cells had a slower growth rate and were arrested at the G0 phase by the suppression of PSMD2. On the other hand, metformin decreased the expression of CAP1 and STIP1, which regulated the cell motility, maintained the cell shape, and reduced the migration of AGS cells. Therefore, we consider that metformin has high potential in new anticancer therapy.

Experimental Section

Chemicals

Metformin (1,1-dimethylbiguanide hydrochloride), triethylammonium bicarbonate buffer (TEABC), thiazolyl blue tetrazolium bromide (MTT), tris(2-carboxyethyl) phosphine hydrochloride (TCEP), S-methylmethanethiosulfonate, Tween 20, and Triton X-100 were purchased from Sigma-Aldrich (St. Louis, MO, USA). Acetonitrile (ACN) was purchased from Lab-Scan. Sequencing grade modified trypsin was purchased from Promega (Madison, WI, USA). Dimethyl sulfoxide was purchased from Scharlau Chemie (Barcelona, Spain).

Cell Culture

AGS cell line was purchased from ATCC (CRL-1739). Cells were cultured in a complete medium (RPMI-1640 medium supplemented with 10% (v/v) FBS) at 37 °C with 5% CO₂. The growing cells were sub-cultured every few days when it reached 80–90% confluency.

MTT Assay

MTT solution was prepared in phosphate-buffered saline (PBS) at a concentration of 5 mg/mL and sterilized by a 0.22 μm filter. 3,000 cells were seeded into a 96-well plate and incubated for 24 h before treated with 1, 5, 10, 20, and 50 mM metformin for 24, 48, and 72 h. Steps were described in our previous paper.⁵⁴

Colony Formation

250 cells were seeded into each well of six-well plate and incubated for 24 h. After 24 h, the medium was changed to a complete medium or a complete medium with 10 mM metformin for treatment and incubated for 7 days at 37 °C. Steps were described in our previous paper.⁵⁴

Cell Cycle and Apoptosis Analysis by Flow Cytometry

1 × 10⁵ cells were seeded into a 10 cm dish and incubated for 24 h, followed by 24 h of starvation in serum-free medium (RPMI-1640 medium without FBS). 10 mM metformin was added for 24, 48, or 72 h. Steps were described in our previous paper.⁵⁴ For apoptosis analysis, 1 × 10⁵ cells were seeded for 24 h and then treated with 10 mM metformin. After being treated with 10 mM metformin for 72 h, the cells were trypsinized and stained with 100 μL binding buffer with annexin V-FITC and PI at room temperature for 15 min. The samples were analyzed by BD FACSCanto II.

Transwell Migration Assay

1 × 10⁵ cells were seeded onto a 10 cm dish and incubated for 24 h and then treated with 10 mM metformin. Cells were trypsinized after 48 h of metformin treatment and seeded 5 × 10⁴ of cells into 8 μm Hanging cell culture inserts (Millipore, Billerica, USA) of the transwell. 10 mM metformin in serum-free medium and a complete medium with 10 mM metformin were added to the bottom of the transwell. The migratory cells were fixed in methanol and stained by crystal violet after 8 h.

Immunofluorescence Staining

1 × 10⁴ cells were seeded on the coverslips in the six-well plate and incubated for 24 h and then treated with 10 mM metformin for 48 h. The coverslip was fixed in 4% (v/v) paraformaldehyde for 15 min at room temperature and permeabilized by 0.25% (v/v) Triton X-100 for 5 min. The coverslip was blocked for 30 min in a blocking solution [1% (w/v) bovine serum albumin (Bioshop) in PBS] at room temperature. Primary mouse anti-human vinculin monoclonal antibody (1:200) (Millipore) and rabbit anti-human ki67 polyclonal antibody (1:100) (Abcam, Cambridge, MA, USA) were diluted in the blocking solution and incubated overnight at 4 °C. After washing twice with PBST (PBS with 0.05% (v/v) Tween-20), samples labeled with vinculin and ki67 were incubated in the secondary anti-mouse IgG FITC antibody (1:100 in PBS; Invitrogen, Carlsbad, CA) and anti-rabbit IgG FITC antibody (1:100 in PBS; Invitrogen, Carlsbad, CA), respectively, with TRITC–phalloidin (1:1250 in PBS; Millipore) for 30 min at room temperature, followed by washing three times with PBST. Subsequently, coverslips were mounted with a ProLong Gold antifade reagent with DAPI (Invitrogen) on the slide. Steps were described in our previous paper.⁵⁴

xCELLigence RTCA Dual-Plate System

To determine whether the inhibition of cell proliferation by metformin is irreversible or reversible, 3,000 cells were seeded into each well of E-plate after 48 h of metformin treatment. Cells were incubated with or without 10 mM metformin and the cell growth was recorded once an hour for 72 h. A detailed protocol was published previously.⁵⁴ To evaluate the cell migration ability after metformin treatment, we use the transwell migration assay and the RTCA DP system with cell invasion/migration (CIM)-plate.

iTRAQ Technique

10 mM metformin-treated cells or control were trypsinized after 72 h and lysis buffer containing 1% (v/v) SDS, 50 mM Tris-HCl, 10% (v/v) glycerol, and protease inhibitor (Bioman, Taipei, Taiwan) were added. The samples were homogenized by a LABSONIC M ultrasonic homogenizer (Sartorius AG, Goettingen, Germany) for 1.5 min on ice and then centrifuged at 17,000×g for 30 min at 4 °C. The proteins in the supernatant were collected with new tubes and quantified with a BCA Protein Assay Reagent kit (Pierce, Rockford, IL, USA). Samples were reduced by TCEP (Sigma-Aldrich) at 37 °C for 30 min and alkylated by iodoacetamide at room temperature in the dark for 30 min.

60 μg of proteins of each sample were adjusted to an equal volume. Proteins were dissolved in 50 mM TEABC to adjust the pH to about 8.5. Subsequently, to the protein solutions were added 5 mM TECP for 30 min at 37 °C and then added 2 mM MTTS at room temperature in the dark for 30 min to reduce and alkylate the cysteines, which might interfere with the iTRAQ signal. For gel-assisted digestion, the protein solution was mixed with acrylamide/bisacrylamide (40%, 37.5:1; Bioshop), APS, and TEMED in the ratio 14:5:0.3:0.3 (v/v) at room temperature for about 10 min of solidifying. The gels were cut into small pieces and washed with 25 mM TEABC and 25 mM TEABC containing 50% (v/v) ACN until no bubble was observed. The gels were dehydrated with 100% ACN and dried completely with a Centrifugal Evaporator CVE-2000 with Uni Trap UT-1000 (Eyela, Japan) for 20 min. The gels were digested with trypsin in 25 mM TEABC (protein amount/trypsin amount = 10:1) in a 37°C water bath overnight. After trypsin digestion, peptides were extracted with 0.1% (v/v) trifluoroacetyl (TFA), 50% (v/v) ACN containing 0.1% (v/v) TFA, and 100% ACN, and peptide solutions were combined and dried with the Centrifugal Evaporator CVE-2000 with Uni Trap UT-1000.

The iTRAQ Reagent kit was purchased from Applied Biosystems (Forster City, CA, USA). The dried peptides were dissolved in iTRAQ dissolution buffer and quantified with a BCA Protein Assay Reagent kit. 30 μg of peptides of each sample were labeled with the isobaric chemical tag. For duplication, the control samples were labeled with iTRAQ114 and iTRAQ115, and the metformin-treated samples were labeled with iTRAQ116 and iTRAQ117 for 1 h at room temperature. All labeled peptides were pooled and dried with the Centrifugal Evaporator CVE-2000 with Uni Trap UT-1000.

Peptides were desalted using ZipTip Pipette Tips (Millipore) before LC–MS/MS spectrometry. The peptides were dissolved in 0.1% (v/v) TFA and adjusted pH to 2–3 with 10% (v/v) TFA. The ZipTip was rinsed with 50% (v/v) ACN containing 0.1% (v/v) TFA by aspirating and dispensing the solution several times. The ZipTip was equilibrated in 0.1% TFA and bound the peptides by aspirating and dispensing the peptide solution 20 times. The peptides in ZipTip were washed with 0.1% TFA and ultimately eluted with 50% ACN containing 0.1% TFA at least 10 times. The desalted peptide solution was dried with the Centrifugal Evaporator CVE-2000 with Uni Trap UT-1000. After desalting, samples were analyzed by LC-ESI-Q-TOF mass spectrometry (Waters SYNAPT G2 HDMS; Waters Corp., Milford, MA, USA) equipped with a nanoACQUITY UPLC system (Waters Corp., Milford, MA, USA). Peptide samples were loaded onto a 2 cm × 180 μm capillary trap column and separated in a 75 μm × 25 cm nanoACQUITY 1.7 μm BEH C18 column. The mobile phases consisted of buffer A (0.1% formic acid) and buffer B (0.1% formic acid in acetonitrile). Peptides were eluted with a linear gradient of 6–50% buffer B. The flow rate was 300 nL/min for 100 min. A NanoLockSpray source was used for accurate mass measurement, and the lock mass channel was sampled every 30 s. A mass spectrometer was calibrated with a synthetic human [Glu¹]–fibrinopeptide B solution (1 pmol/μL; Sigma-Aldrich) delivered through the NanoLockSpray source. Data acquisition was performed using the data-directed analysis (DDA). The DDA method included one full MS scan (m/z 350–1700, 1 s) and three MS/MS scans (m/z 100–1990, 1.5 s for each scan) sequentially on the three most intense ions present in the full-scan mass spectrum. Each sample was analyzed in technical replicates. Mass spectral data were deposited at 109 ProteomeXChange (http://www.proteomexchange.org/), project accession number PXD024996. Details for the protocols were published previously.³²

Protein Identification

Mass spectral data were converted to mgf and mzXML file formats using Mascot Distiller (version 2.3.2; Matrix Science, London, United Kingdom) and massWolf (version 4.3.1; Institute for Systems Biology, Seattle, WA, USA), with default settings, respectively. The mgf files were submitted to Mascot Daemon (version 2.3.2; Matrix Science) for peptide identification. For peptide identification, the MS/MS spectral data were searched against the Homo sapiens of SwissProt database with the following parameters: trypsin defined as an enzyme with two missed cleavages allowed, a tolerance of 0.1 Da on the peptide ion, and an MS/MS fragment ion and variable modification settings such as iTRAQ4plex (N-term), iTRAQ4plex (K), deamidated (NQ), oxidation (M), and methylthio (C). Mascot search results were exported to the XML file format after being filtered by the significance threshold p < 0.05 and the ion score cutoff of 0.05. The XML and mzXML files were submitted to Multi-Q software (version 1.6.5.4)⁵⁵ for selecting qualified peptides. The peptides that satisfied the following criteria were selected for further analysis: the peptide is labeled with iTRAQ; the peptide is nondegenerate; the ion score of the peptide is higher than the Mascot identity score (P < 0.05); and the average of all iTRAQ intensities higher or equal to 30.

Survival Analysis

The Kaplan–Meier overall survival analysis was performed for the gastric cancer patient with Gene Expression Omnibus accession number GSE15459.⁵⁶

Statistical Analysis

The log-rank (Mantel–Haenszel) test was performed by R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl). P-value <0.05 means two groups have statistical significance.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.1c00894.

• Metformin does not result in apoptosis but affects the cell cycle and migration in AGS cells (PDF)

This research was funded by the Ministry of Science and Technology (MOST 109-2221-E-010-012-MY3, MOST 109-2320-B-002-017-MY3, and MOST 109-2221-E-010-011-MY3), the Higher Education Sprout Project (109L8837A), and the National Health Research Institutes (NHRI-EX109-10709BI) in Taiwan. The APC was funded by MOST 109-2320-B-002-017-MY3.

The authors declare no competing financial interest.