A Comparison Study on the Metabolites in PC-3, RWPE-1, and Chrysin-Treated PC-3 Cells

Abstract

Prostate cancer is frequently diagnosed and the leading cause of death in men worldwide. Prostate-specific antigen (PSA) blood tests and biopsies are the primary methods for diagnosing prostate cancer; however, their accuracy is less than 50%. Therefore, there is a need to develop diagnostic tests that minimize patient discomfort during examination and adequate biomarkers that are more accurate, sensitive, and specific for the detection of prostate cancer. This study investigated the application of metabolomics to identify biomarkers in prostate cancer biofluids. In addition, changes in prostate cancer metabolite levels induced by chrysin, a natural anticancer compound, were evaluated and compared with those in non-treated prostate cancer cells. Gas chromatography-mass spectrometry (GC-MS)-based metabolomic profiling was performed to investigate the differences in metabolic alterations among prostate cancer, normal prostate, and chrysin-treated prostate cancer cells. Pairwise comparisons of the extracellular fluid metabolomes were performed using principal component analysis (PCA), partial least squares–discriminant analysis (PLS-DA), and Student’s t-test. The results revealed significantly different patterns among the metabolite groups, including alcohols, amino acids, carboxylic acids, organic acids, sugars, and urea. The RWPE-1- and chrysin-treated PC-3 (PC-3 Chr) cell groups showed similar tendencies for 23 metabolites, while the groups showed significant differences from the PC-3 group. Most amino acids showed higher concentrations in PC-3 cells than in the normal cell line RWPE-1 cells and PC-3 Chr cells. Our results revealed that GC-MS might be an effective diagnostic tool to detect prostate cancer and contribute to finding new tumor markers for prostate cancer as the basis for new ideas.

1. Introduction

Prostate cancer is a threat to human health worldwide. In 2020, prostate cancer was the second most commonly diagnosed cancer and the fifth leading cause of cancer-related deaths among men worldwide in 2020 [1].

Prostate cancer screening tests and diagnoses include digital rectal examination, prostate-specific antigen (PSA) testing, ultrasound, magnetic resonance imaging (MRI), and prostate biopsy [2]. Except for the PSA test, these tests cause discomfort to the patient, as they involve examining the rectum. The PSA test is mainly used for cancer detection through blood tests because PSA is produced in the prostate and semen, and small amounts of PSA circulate in the blood. Prostate cancer can cause elevated PSA levels, but many non-cancerous conditions can also increase PSA levels. Therefore, the PSA level does not provide precise diagnostic information, making the diagnosis difficult using the PSA test alone. Thus, there is a need to develop diagnostic tests that minimize patient discomfort during examination and adequate biomarkers that are more accurate, sensitive, and specific for the detection of prostate cancer.

Metabolomics is a quantitative study that measures low-molecular-weight metabolites such as nutrients, drugs, and signaling mediators in an organism at a specified time under specific environmental conditions [3]. Recently, metabolomics has received increased attention for its application in early cancer diagnosis [4] because the occurrence and development of cancer can cause changes to the metabolome of the human body. Metabolomics is a powerful tool for identifying cancer biomarkers and improving cancer diagnosis, monitoring, and treatment [5]. Metabolomic studies generally employ nuclear magnetic resonance (NMR) spectroscopy, high-performance liquid chromatography/mass spectrometry (HPLC/MS), Fourier-transform infrared (FT/IR) spectroscopy, and gas chromatography–mass spectrometry (GC/MS) [6].

Anticancer effects such as apoptotic effect [7], anti-inflammatory effect [8], anti-proliferative effect [9], and target of cancer-promoting transcription factors [10] of chrysin have been reported in various cancers, including prostate cancer, ovarian cancer, and breast cancer. In this study, chrysin was evaluated for metabolomic applications in cancer treatment monitoring. Metabolic differences between normal prostate, prostate cancer, and chrysin-treated PC-3 (PC-3 Chr) cells were analyzed using GC-MS to identify prostate cancer-specific metabolites and evaluate the monitoring of cancer treatment.

Exploring alterations in metabolite profiles following treatment with anticancer agents can reveal potential biomarkers for assessing drug efficacy and responses. In this study, we compared the metabolic profiles of PC-3 cells, a widely used prostate cancer cell line, and PC-3 Chr cells. We evaluated the differences in metabolites between the untreated and PC-3 Chr cells to elucidate the metabolic alterations induced by this therapeutic intervention.

We also analyzed metabolites in benign prostatic hyperplasia (BPH) cells. BPH, a non-cancerous condition characterized by increased PSA levels, served as an essential control in our study. Comparison of the metabolite profiles of PC-3 cells, PC-3 Chr cells, and BPH cells enabled us to discern specific metabolic changes associated with prostate cancer and distinguish them from those occurring in benign conditions.

2. Materials and Methods

2.1. Test Chemical

Methoxyamine hydrochloride, N,O-bis trimethylsilyl trifluoroacetamide (BSTFA), and pyridine (anhydrous 99.8%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Fluoranthene was purchased from Supelco (Bellefonte, PA, USA). Chrysin (M.W. = 254.241 g/mol, purity ≥ 97% as determined through HPLC) was purchased from Sigma-Aldrich (Cat: C80105, St. Louis, MO, USA).

2.2. Cell Culture and GC-MS

PC-3 cells (human castration-resistant cancer cell line No. 21435, Korean Cell Line Bank, Seoul, Republic of Korea) and RWPE-1 cells (benign prostate cell line; ATCC, Manassas, VA, USA) were cultured in RPMI 1640 (Cat: LM 011-01, Welgene, Daegu, Republic of Korea) medium supplemented with 10% fetal bovine serum (FBS) (Cat: S101-07, Welgene, Daegu, Republic of Korea) and 1% antibiotics (Cat: LS203-01, Welgene, Daegu, Republic of Korea) in a 5% CO₂ incubator at 37 °C. PC-3 and RWPE-1 cells (3 × 10⁵ cells) were cultured in 6 wells for 24 h. PC-3 cells (3 × 10⁵ cells) were treated with 10 μM chrysin for 24 h. The cell supernatants were collected for metabolomic profiling using untargeted GC-MS. Untargeted GC-MS-based metabolomic profiling was performed by the National Instrumentation Center for Environmental Management in Seoul National University (Seoul, Republic of Korea). The details of this method are as follows: Approximately 20 mg of the dried cell culture medium was mixed with 1 mL of 70% MeOH by sonication for 30 min. After filtration with 0.2 μm microporous membranes, 150 μL of the collected supernatant was transferred into a GC vial and dried using a speed vac. For the analysis of compounds by GC-MS, dried extract samples were redissolved and derivatized by addition of 50 μL of 20 mg/mL methoxyamine hydrochloride in pyridine and incubated at 30 °C for 90 min for oximation. Then, 50 μL of BSTFA and 30 uL of fluoranthene (1000 ug/mL in pyridine as an internal standard) were added to each sample for trimethylsilylation derivatization, and the mixture was heated at 60 °C for 30 min. The samples were then subjected to GC-MS. Each 1.0 μL aliquot of derivatized sample was injected in a 30:1 split ratio into an Thermo Scientific (Trace 1310/ISQ LT, Seoul, Republic of Korea) GC-MS equipped with DB-5MS capillary column (60 m × 0.25 mm × 0.25 μm, Agilent J&W Scientific, Santa Clara, CA, USA). Helium was used as the carrier gas at a constant flow rate of 1.5 mL/min. The temperature program was as follows: The initial temperature was 50 °C held for 2 min, raised to 180 °C at a rate of 5 °C/min and held for 8 min, increased to 210 °C at a rate of 2.5 °C/min, elevated to 325 °C at a rate of 5 °C/min, and maintained for 10 min. The temperature of the injector, transfer line, and ion source were set to 300, 320, and 270 °C, respectively. A mass range (35–680 m/z) in full-scan mode for electron impact ionization (70 eV) was applied. The solvent delay time was set to 14 min.

2.3. Identification and Quantification of Metabolites

The identification of each metabolite was confirmed by comparing its retention time and mass spectral data with those of the NIST/EPA/NIH Mass Spectral Library (version 2.0 d, National Institute of Standards and Technology, Gaithersburg, ML, USA). All metabolites were identified by comparing mass fragments with standard mass spectra in the NIST (National Institute of Standards and Technology) commercial database with a similarity of more than 70%. In addition, Kovat retention index (RI) information (i.e., the retention times of the alkanes) was obtained by direct injection of the standard solution under the same GC-MS conditions. For the RI calculation, a standard solution containing C8–C30 normal alkanes in dichloromethane was used. The area of the corresponding peak was calculated by integrating the peak intensity of the selected ion monitoring chromatogram with the total ion chromatogram (the parameters are shown in

Table 1. Metabolomics profiling by GC-MS.

Compounds	RT 1 (min)	RWPE-1 2	PC3	PC3_C	Quantitative Ion	TMS 3	RI 4
Alcohols (6)
Threitol	28.36	31.63	37.27	34.02	147	4	1494
Arabinitol	35.42	2.92	3.87	2.84	103	5	1706
Arabitol	35.66	1.95	2.91	2.36	103	5	1711
Mannitol (1)	45.13	2.44	2.91	2.36	147	6	1918
Mannitol (2)	45.43	38.93	55.18	46.76	147	6	1925
Myoinositol	51.53	103.67	172.86	155.84	305	6	2080
Amino acids (7)
Valine	16.87	22.43	32.95	N.D.	72	1	1085
Leucine	18.93	9.28	58.63	N.D.	86	1	1152
Isoleucine (1)	19.58	24.89	73.19	N.D.	86	1	1174
Proline	19.66	N.D. 5	78.04	N.D.	70	1	1177
Isoleucine (2)	22.81	N.D.	92.82	N.D.	158	2	1285
Glycine	23.20	N.D.	113.34	N.D.	174	3	1299
Pyroglutamic acid	28.93	144.13	637.21	13.40	156	2	1515
Carboxylic acids (7)
Propionic acid	15.56	11.68	23.74	5.66	174	1	1043
2-Hydroxybutyric acid	17.91	9.73	4.84	4.25	131	2	1119
Pentanoic acid	19.51	8.28	6.78	N.D.	89	1	1172
2-Keto-3-methylpentanoic acid	20.11	0.97	2.41	N.D.	89	1	1192
2-Ketoisocaproic acid	20.54	2.92	4.84	N.D.	89	1	1206
2-Pyrrolidone-5-carboxylic acid	28.48	167.95	8.24	159.85	84	1	1499
2,3,4-Trihydroxybutyric acid	29.30	3.41	8.23	8.53	147	4	1528
Organic acids (5)
Lactic acid	15.86	11,412.12	13,860.03	13,449.80	147	2	1052
Oxalic acid	18.37	9.73	43.08	N.D.	147	2	1134
Succinic acid	23.43	40.90	53.30	36.89	147	2	1307
Malic acid	27.94	4.38	8.24	4.73	147	3	1478
2-Hydroxyglutaric acid	30.29	0.97	12.59	N.D.	129	3	1562
Sugar acid (1)
Glyceric acid	23.78	0.97	0.97	6.15	189	3	1320
Sugars (7)
Tagatofuranose	39.42	2.42	20.18	0.48	217	5	1972
Fructopyranose	39.98	2.43	35.23	N.D.	204	5	1804
Fructose (1)	42.58	255.01	250.75	245.14	103	5	1861
Fructose (2)	43.02	182.02	165.35	191.22	103	5	1871
Glucose (1)	43.69	0.48	192.41	170.02	319	5	1885
Glucose (2)	44.54	N.D.	24.22	20.78	319	5	1904
Glucose (3)	47.49	N.D.	86.82	4.74	204	5	1972
Others (2)
Urea	21.37	N.D.	210.52	N.D.	147	2	1235
Phosphoric acid	22.18	996.83	1210.09	1281.87	299	3	1263

¹ RT, retention time; ² Concentration, fluoranthene equivalent ug/mg, mean values (n = 2); ³ TMS, trimethylsilylation; ⁴ RI, retention index; ⁵ N.D., not detected.

). The calculated area of each compound was normalized by dividing it by the internal compound (fluoranthene) peak area to obtain a semi-quantitative composition of the components. Each compound was quantified against an internal standard by integrating peaks.

2.4. Statistical Data Analysis

All analyses were performed in duplicates. To verify the separation trends between different groups, partial least squares–discriminant analysis (PLS-DA) was conducted using the SIMCA software (version 17, Umetrics, Umea, Sweden). A PLS-DA model was therefore constructed for classification of samples variables to determine the important compounds among samples. The data were expressed as means ± standard deviation (SD) of three replicates per experiment using Graph Pad Prism software 8.0 (GraphPad Software, Boston, MA, USA). One-way analysis of variance (ANOVA) was used to assess the significance of the differences between groups. Statistical significance was set at p < 0.05.

3. Results

3.1. Profiling of Metabolites

In total, 35 metabolites were identified in the GC-MS datasets obtained from the cells. Table 1 shows these compounds, including six alcohols (threitol, arabinitol, arabitol, two kinds of mannitol, and myoinositol), seven amino acids (valine, leucine, two kinds of isoleucine, proline, glycine, and pyroglutamic acid), seven carboxylic acids (propionic acid, 2-hydroxybutyric acid, pentanoic acid, 2-keto-3-methylpentanoicacid, 2-ketoisocaproic acid, 2-pyrrolidone-5-carboxylicacid, and 2,3,4-trihydroxybutyric acid), five organic acids (lactic acid, oxalic acid, succinic acid, malic acid, and 2-hydroxyglutaric acid), one sugar acid (glyceric acid), seven sugars (tagatofuranose, fructopyranose, two kinds of fructose, and three kinds of glucose), and two others (urea and phosphoric acid).

3.2. Multivariate Analysis

PLS-DA was performed to clearly differentiate metabolic differences in the samples. Figure 1 presents the PLS-DA-derived score plot based on the first (64%) and second (23.6%) principal PLS components. The PLS-DA score plot indicated a separation between group 2 and groups 1 and 3. The parameters of the cross-validation model were component 3, with R2X = 0.955, R2Y = 0.994, and Q2 (cum) = 0.975. The R2Y and Q2Y demonstrated the robustness of the PLS-DA model. R2Y describes how well the training set data are mathematically reproduced and varies between 0 and 1, with 1 indicating a perfect fit of the model. Q2Y quantifies the ability of the model to reliably predict the outcome of other experiments, with Q2Y values >0.5 being considered to have good predictability [11]. The main metabolites were identified based on a variable importance plot (VIP) values of 1.0. Among the identified metabolites from cell samples, the highest VIP values (1.40–1.02) came from pentanoic acid (1.40), 2-hydroxybutyric acid (1.38), glyceric acid (1.33), glucose (1.29), 2,3,4-trihydroxybutyric acid (1.26), myoinositol (1.20), lactic acid (1.19), phosphoric acid (1.16), valine (1.12), 2-ketoisocaproic acid (1.12), and mannitol (1.02). The correlation matrix shows the pairwise correlation between all variables (X and Y) in the current workset, scaled and transformed into the workset. Each variable is displayed in one row and one column in the correlation matrix, and the correlation between two variables is shown in the cell where the two variables intersect. When the correlation coefficient was close to zero, there was no linear relationship between the terms. The major metabolite correlation matrix for RWPE-1 was hydroxybutyric acid, whereas that for PC-3 was arabinitol and glycine (Figure 2b). The major metabolite in the correlation matrix of the chrysin-treated PC-3 cells was glyceric acid.

3.3. Metabolic Difference between Groups RWPE-1 and PC-3 Cells

As shown in Figure 2, 18 differentially expressed metabolites were identified, mainly two types of alcohols: arabinitol and mannitol; five amino acids: leucine, proline, isoleucine, glycine, and pyroglutamic acid; four carboxylic acids: propionic acid, 2-keto-3-methyl pentanoic acid, 2-keto-isocaproic acid, and 2-pyrrolidone-5-carboxylic acid; four organic acids: oxalic acid, succinic acid, malic acid, and 2-hydroxyglutaric acid; three sugars: tagatofuranose, fructopyranose, and glucose; and urea. Differences in the metabolites were compared between groups 1 (RWPE-1 cells) and 2 (PC-3 cells). The PC-3 cell group showed increased concentrations of all metabolites except 2-pyrrolidone-5-carboxylic acid compared to the normal RWPE-1 cell group. Leucine, proline, isoleucine, glycine, and pyroglutamic acid showed 6.3-, 78-, 92.8-, 113.3-, and 4.4-fold higher concentrations, respectively, in PC-3 cells than in RWPE-1 cells, whereas 2-pyrrolidone-5-carboxylic acid presented a 20-fold lower concentration in PC-3 cells (Figure 2b). No glucose or urea was detected in the RWPE-1 group (Figure 2e).

3.4. Metabolic Difference between Groups PC-3 and Chrysin-Treated PC-3 Cells

To determine whether metabolomic profiling could prove the effect of anticancer drugs and target metabolites of chrysin, chrysin, a well-known anticancer natural compound, was applied as treatment to PC-3 cells, and the differences in metabolites were compared with those in the PC-3 cell group. As shown in Figure 3, 15 significantly different metabolites were identified, including six amino acids: leucine, two types of isoleucine, proline, glycine, and pyroglutamic acid; four carboxyl acids: propionic acid, 2-keto-3-methyl pentanoic acid, 2-keto-isocaproic acid, and 2-pyrrolidone-5-carboxylic acid; four organic acids: oxalic acid, succinic acid, malic acid, and 2-hydroxyglutaric acid; sugars; and others: fructopyranose and urea. The PC-3 Chr group showed a decrease in all metabolite concentrations except 2-pyrrolidone-5-carboxylic acid compared to cancer cells in the PC-3 group. The PC-3 group showed an opposite trend to that of the PC-3 Chr group. Leucine, proline, isoleucine, and glycine were not detected, whereas the concentration of 2-pyrrolidone-5-carboxylic acid was 19.5-fold higher in the PC-3 Chr group than in the PC-3 group (Figure 3). Furthermore, 2-keto-3-methyl pentanoic acid, 2-keto-isocaproic acid, oxalic acid, 2-hydroxyglutaric acid, fructopyranose, and urea were not detected in the PC-3 Chr group. In contrast, the levels of these metabolites were increased in PC-3 group (Figure 3).

3.5. Metabolic Difference between Groups RWPE-1, PC-3, and Chrysin-Treated PC-3 Cells

The chrysin-treated PC-3 cell group was similar to the RWPE-1 group in 23 (e.g., arabinitol, arabitol, mannitol, myoinositol, valine, leucine, isoleucine(1), isoleucine, proline, glycine, pyroglutamic acid, propionic acid, 2-keto-3-methylpentanoic acid, 2-ketoisocarproic acid, 2-pyrrolidone-5-carboxylic acid, oxalic acid, succinic acid, malic acid, 2-hydorxyglutaric acid, tagatofuranose, fructopyranose, glucose, and urea) of 35 metabolites. Therefore, when considering the results, the PC-3 group was distinctively different from the RWPE-1 and chrysin-treated PC-3 groups.

4. Discussion

This study investigated the metabolome as a valuable source of prostate cancer biomarkers by distinguishing between normal prostate and prostate cancer cells. Additionally, metabolomic applications for monitoring cancer treatment were explored by comparing the metabolome of PC-3 cells with that of chrysin-treated cells by GC-MS analysis.

GC-MS-based metabolomic analysis in cancer research is an important platform for evaluating cancer development through metabolic profiling and earlier identification of more sensitive diagnostic makers [12,13,14,15]. GC-MS-based metabolomic analysis has been developed for metabolite extraction from cell cultures, tissues, and serum/plasma because GC-MS can analyze water-soluble cellular materials, water-insoluble fatty acids, and phospholipids [16].

This study found significant differences in the metabolite profiles between normal prostate cells and prostate cancer cells. As shown in Figure 2, the average levels of 18 metabolites differed significantly between the normal prostate and prostate cancer cell groups. In addition, PC-3 Chr cells showed a difference in metabolic profiles from PC-3 cells (Figure 3 and Figure 4), whereas the PC-3 Chr cell group was similar to the RWPE-1 cell group in the metabolome. Chrysin has been shown to exert anticancer activities through emerging cellular and molecular mechanisms, including the inhibition of invasion and metastasis [17], oxidative stress [18], inflammation, angiogenesis [19], proliferation, and induction of apoptosis [9] in prostate cancers [20]. Recently, we showed that chrysin has anticancer effects, including the inhibition of epithelial–mesenchymal transition (EMT), vasculogenic mimicry, angiogenesis, and the induction of apoptosis by inhibiting the SPHK/HIF-1alpha signaling pathway in hypoxia-induced PC-3 cells [10]. Chrysin, which has anticancer effects, was used to evaluate the application of GC-MS-based metabolic analysis for monitoring cancer treatment. This study demonstrated the metabolic changes induced by chrysin in cancer cells.

This study showed that the average levels of amino acids, alcohols, carboxylic acids, organic acids, sugars, and urea were significantly different in cancer and normal cells. Furthermore, multivariate analysis and comparison of metabolite concentration levels showed that the normal and chrysin-treated cell groups were highly similar and differed significantly from the PC-3 group. Predominantly, proline, isoleucine, and glycine amino acids and urea were not detected in RWPE-1 and PC-3 Chr cells, while these amino acids were highly detected in the PC-3 cell group. Similar to our study, a pilot clinical study showed that the amino acid profiles of serum and urine might provide valuable and clinically useful markers for prostate cancer [21]. Glycine is essential for the synthesis of proteins, nucleic acids, and lipids that are important for cell growth and tumor homeostasis [22]. Furthermore, glycine metabolism correlates with cell proliferation and poor prognosis in several tumors [23]. Proline metabolism triggers with conversion into Δ1- pyrroline-5-carboxylate (P5C) by proline dehydrogenase enzyme (PRODH) in tumor growth and metastatic progression. PRODH enzymes, involved in proline synthesis and catabolism, were reported to induce PRODH expression under hypoxic conditions in different breast cancer cell lines, and a mouse xenograft model was shown to contribute to cancer cell survival [24]. An increase in urea cycle metabolites is characteristic of prostate cancer. Several studies have reported urea cycle metabolites, including aspartate, arginosuccinate, arginine, proline, and the oncometabolite fumarate, are higher in prostate cancer than in benign controls, and fumarate levels correlate positively with HIF-1α and NF-κB pathways [25]. Additionally, high urea concentrations in the serum or urine are linked to the risk of prostate cancer [26]. Changes in the aforementioned metabolites (e.g., proline, isoleucine, glycine, and urea) by chrysin may be due to the inhibitory effect of chrysin on HIF-1α.

5. Conclusions

GC-MS-based metabolome data revealed that the normal and chrysin-treated cell groups were highly similar and differed significantly from the PC-3 group. The RWPE-1 and PC-3 Chr cell groups showed similar tendencies for the 23 metabolites. Furthermore, amino acids and urea were shown to be usable as potential biomarkers, and metabolomic studies of prostate cancer using GC-MS-based metabolomic analysis can help identify cancer markers and monitor cancer treatment.