Proteomic and computational analyses followed by functional validation of protective effects of trigonelline against calcium oxalate-induced renal cell deteriorations

Abstract

Trigonelline is a phytoalkaloid commonly found in green and roasted coffee beans. It is also found in decaffeinated coffee. Previous report has shown that extract from trigonelline-rich plant exhibits anti-lithiatic effects in a nephrolithiatic rat model. Nevertheless, cellular mechanisms underlying the anti-lithiatic properties of trigonelline remain hazy. Herein, we used nanoLC-ESI-Qq-TOF MS/MS and MaxQuant-based quantitative proteomics to identify trigonelline-induced changes in protein expression in MDCK renal cells. From a total of 1006 and 1011 proteins identified from control and trigonelline-treated cells, respectively, levels of 62 (23 upregulated and 39 downregulated) proteins were significantly changed by trigonelline. Functional enrichment and reactome pathway analyses suggested that these 62 altered proteins were related to stress response, cell cycle and cell polarity. Functional validation by corresponding experimental assays revealed that trigonelline prevented calcium oxalate monohydrate crystal-induced renal cell deteriorations by inhibiting crystal-induced overproduction of intracellular reactive oxygen species, G0/G1 to G2/M cell cycle shift, tight junction disruption, and epithelial-mesenchymal transition. These findings provide cellular mechanisms and convincing evidence for the renoprotective effects of trigonelline, particularly in kidney stone prevention.

Graphical Abstract

Highlights

• •Trigonelline induces altered levels of 62 proteins in MDCK renal cells.
• •The altered proteins are related to stress response, cell cycle and cell polarity.
• •COM crystals induce oxidative stress and cell cycle shift from G0/G1 to G2/M.
• •COM crystals disrupt tight junction and cause epithelial-mesenchymal transition.
• •All of these COM-induced defects can be prevented by trigonelline pretreatment.

1. Introduction

Trigonelline is a bioactive phytoalkaloid predominantly found in a traditional herb namely fenugreek or Trigonella foenum-graecum [1], [2]. This compound is also found in green and roasted coffee beans [3]. Arabica green coffee contains trigonelline at 9.87 – 12.60 mg/g [4], whereas Arabica expresso and Robusta coffee contain trigonelline at 70 mg/cup and 40 mg/cup, respectively [5]. Roasting condition also affects trigonelline content, e.g., fast-roasting process better preserves trigonelline content than slow-roasting procedure [6], [7]. Roasting at temperature < 225 ℃ is recommended to enrich trigonelline content because the higher roasting temperature causes trigonelline degradation [3]. Trigonelline content in the coffee roasted at 225 ℃ for 120 s is approximately 0.80 g/100 g dry weight [3]. Interestingly, trigonelline is not much affected by decaffeinated process, as it is slightly reduced from 0.80 to 0.69 g/100 g dry weight after decaffeination [3].

Serum and urinary trigonelline levels can serve as biomarkers for coffee consumption [8], [9], [10]. The potential benefits of trigonelline have been revealed in several diseases, such as Alzheimer’s disease [11], [12], [13] and diabetes [11]. Kidney stone disease (also known as nephrolithiasis/urolithiasis) is a result of deposition of crystalline compounds packed as the solid mass within the kidney. Among the crystalline compounds, calcium oxalate monohydrate (COM) is the predominant type found in clinical stones [14], [15]. The in vivo anti-lithiatic effects of extract from trigonelline-rich plant have previously been reported [16]. A recent study has demonstrated the direct effects of trigonelline on COM crystals in kidney stone forming mechanism, including reduction of COM crystal size/number during crystallization, inhibition of COM crystal growth, and prevention of adherence of the crystals on cellular surfaces [17].

Renal epithelial cells form a monolayer to line along renal tubules. The epithelial cell sheet acts as a barrier for regulating fluid and electrolyte homeostasis as well as urine production. Defects in renal epithelial cells contribute to several kidney diseases, including chronic kidney disease (CKD) [18], renal fibrosis [19], diabetic nephropathy [20], and kidney stone disease [21]. Injury of the renal epithelial cell sheet can be induced by several factors, e.g., physical forces [22], drugs [23], high-glucose condition [24], chemicals [25], and crystals [26], [27]. During the past decade, the use of bioactive compounds for relieving and protecting renal epithelial cell injury has gained wide attention [28], [29], [30]. However, precise protective cellular mechanisms and response of renal cells to trigonelline remain hazy.

Therefore, we aimed to explore cellular mechanisms underlying the anti-lithiatic effects of trigonelline. Renal epithelial cells were treated with physiologic (subtoxic) concentration of trigonelline and subjected to quantitative proteomics to assess altered cellular proteome in response to trigonelline. Functional enrichment and reactome pathway analyses were performed to determine relevant biological pathways associated with the trigonelline-induced altered proteins. Finally, corresponding functional assays were performed to validate such relevant biological pathways related to the protective cellular mechanisms and the anti-lithiatic effects of trigonelline.

2. Materials and Methods

2.1. Cultivation of renal cells

MDCK (Madin-Darby canine kidney) cells (ATCC; Manassas, VA) were grown in a growth medium containing DMEM (Gibco; Grand Island, NY), antibiotics (penicillin G at 60 U/ml and streptomycin at 60 μg/ml) (Sigma-Aldrich; St. Louis, MO) and 10% fetal bovine serum. The cells were maintained in a CO₂ incubator (Thermo Fisher Scientific; Marietta, OH) at 37 °C with 5% CO₂ and relative humidity > 95%.

2.2. Trigonelline treatment

Approximately 1 × 10⁶ cells were suspended in 2 ml growth medium and seeded into 6-well culture plate (Corning Inc.; Corning, NY). After 24-h incubation, the growth medium was refreshed and added with 1, 10 or 100 µM trigonelline (Sigma-Aldrich). The cells were further incubated with or without trigonelline for 24-h and then subjected to subsequent experiments as detailed below.

2.3. Cell death evaluation

Trypan blue exclusion assay was performed to select optimal concentration of trigonelline for subsequent experiments. The optimal concentration was set as the maximum that did not significantly induce cell death. After 24-h incubation with/without trigonelline, cell morphology was examined under an Eclipse Ti-S phase-contrast inverted microscope (Nikon; Tokyo, Japan). Floating and adherent cells were harvested and stained with 0.4% trypan blue solution (Gibco). Using a hemacytometer, the stained dead cells and unstained viable cells were counted and percentage of cell death was calculated as follows.

Cell death (%) = (No. of dead cells / No. of viable and dead cells) × 100%

2.4. In-solution tryptic digestion, nanoLC-ESI-Qq-TOF MS/MS analysis, and label-free quantitative proteomics

After 24-h incubation with/without 100 µM trigonelline, cellular proteins were extracted using SDT lysis buffer (4% SDS, 100 mM DTT, and 100 mM Tris-HCl; pH 7.6). Protein concentrations were measured using Bio-Rad protein assay (Bio-Rad; Milano, Italy) based on Bradford’s method. Equal amount (30 µg) of proteins from each sample was used for in-solution tryptic digestion as described previously [31], [32]. The digested peptides were then analyzed by nanoLC-ESI-LTQ-Orbitrap MS/MS as previously reported [33], [34]. The MS raw files (.d) were analyzed using MaxQuant software package (version 1.6.2.6) (https://www.maxquant.org) with built-in Andromeda search engine as described elsewhere [35]. More details of tryptic digestion and mass spectrometric analyses are provided in Supplementary Methods.

For quantitative analysis, label‐free quantification (LFQ) option was enabled with a minimum ratio count of 2. The LFQ intensity (derived from three biological replicates per group and technical triplicate per each biological sample) (generated according to the MaxLFQ algorithm [36]) was used as the quantified value. Peptides with less than five out of nine quantified values in each group were excluded from quantitative analysis. Multivariate statistical analysis was performed to determine the separation of the data derived from each group. The ropls R package [37] was employed to create orthogonal partial least squares discriminant analysis (OPLS-DA) model. The model quality was evaluated with R2X, R2Y (representing good fitness of the model) and Q2Y (representing predictability of the model). To avoid model overfitting, the OPLS-DA model was validated by implementing seven-fold cross validation and applying 200 random permutations. The variable influence on projection (VIP) values indicating the impact of the quantified proteins in the OPLS-DA model were computed. The protein quantification and univariate statistical analysis were carried out using unpaired Student’s t-test. The proteins with p values < 0.05 and VIP values > 1 were considered as significantly altered proteins.

2.5. Western blotting

After 24-h incubation with/without 100 µM trigonelline, cellular proteins were extracted with Laemmli’s buffer and protein concentrations were measured using Bio-Rad protein assay based on Bradford’s method. An equal amount (30 µg) of total proteins from each sample was subjected to Western blotting as described previously [38], [39]. See more details in Supplementary Methods.

2.6. Functional enrichment and reactome pathway analyses

ClueGO plugin (version 2.5.9) (https://apps.cytoscape.org/apps/cluego) and Cytoscape software (version 3.9.1) (https://cytoscape.org/) were used to acquire functional enrichment and reactome biological pathways of all significantly altered proteins induced by trigonelline based on the biological process branch of Gene Ontology (GO).

2.7. Calcium oxalate monohydrate (COM) crystal treatment

COM crystals were produced using the method published previously [40], [41]. After 24-h pretreatment with/without 100 µM trigonelline, the cells were incubated with COM crystals (100 µg crystal/ml medium) for further 24-h. The cells were then subjected to quantitative analysis of intracellular reactive oxygen species (ROS), cell cycle analysis, transepithelial electrical resistance (TEER) measurement, evaluation of zonula occludens-1 (ZO-1) level, and measurements of cell spindle index, epithelial marker (E-cadherin) and mesenchymal marker (vimentin) as detailed below. In parallel, the cells without trigonelline and COM crystal treatments served as the control.

2.8. Quantitative analysis of intracellular ROS

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, the cells were subjected to quantitative analysis of intracellular ROS level using DCFH-DA (dichlorodihydrofluorescein diacetate) assay and flow cytometry as previously described [42], [43]. Briefly, the cells were detached by trypsinization and resuspended in the growth medium followed by 20-min incubation with 5 μM DCFH-DA (Invitrogen-Molecular Probes; Burlington, Canada) at 37 °C in the dark. A flow cytometer (BD Accuri C6) equipped with Accuri C6 software (BD Biosciences; San Jose, CA) was employed to quantify the DCFH-DA-positive cells from at least 10,000 acquisitions in each sample.

2.9. Cell cycle analysis

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, the cells were subjected to flow cytometric analysis of cell cycle distribution as described previously [44], [45]. Briefly, the adherent cells were detached by trypsinization and incubated with ice-cold 70% ethanol for 30-min. After washing with ice-cold PBS, the cells were incubated with 100 μg/ml RNase A (Sigma-Aldrich) in PBS on ice for 30-min. The cells were further stained with propidium iodide (BD Biosciences) at 25 °C for 10-min in the dark. BD Accuri 6 flow cytometer equipped with Accuri C6 software (BD Biosciences) was employed to quantify the cells from at least 10,000 acquisitions in each sample. Cell cycle distribution was then analyzed according to DNA contents of the stained cells.

2.10. TEER measurement

TEER measurement was performed as previously described [46], [47]. Briefly, polyethylene culture inserts (surface area 1.12 cm²) of the 12-mm Transwell plate (0.4-µm pore size) (Corning) were pre-coated with collagen type IV (Sigma-Aldrich). The cells resuspended in the growth medium were seeded onto the upper chamber, whereas the lower chamber was filled with the growth medium. The polarized MDCK cells were subjected to 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals as described above. The cells were then subjected to TEER measurement using Millicell-ERS Voltohmmeter equipped with chopstick electrode (Millipore; Bedford, MA). The TEER values were measured from at least three different sites in each well. The cell-free wells filled only with the growth medium severed as the blank for background subtraction. TEER values were calculated as follows.

TEER (Ω·cm²) = (Resistance _sample – Resistance _blank) (Ω)× culture area (cm²)

2.11. Immunofluorescence staining

Immunofluorescence study was performed as described previously [48], [49]. Briefly, the cells were grown on cover slips and subjected to 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals as described above. The cells were then rinsed with PBS containing 0.1 mM CaCl₂ and 1 mM MgCl₂ and fixed and permeabilized with 4% paraformaldehyde and 0.1% TritonX-100, respectively (15-min each). After extensive wash, they were incubated with 5% bovine serum albumin (BSA) in PBS and then with mouse monoclonal anti-ZO-1 antibody (Invitrogen) (1:100 in 1% BSA/PBS) at 4 °C overnight. After washing, the cells were incubated with Alexa Fluor 488-conjugated corresponding secondary antibody (Invitrogen) (1:10,000 in 1% BSA/PBS) mixed with Hoechst dye (Invitrogen) (1:1000) at 37 °C for 1-h. After washing and mounting on glass slides, the cells were imaged under Eclipse 80i fluorescence microscope equipped with NIS-Elements D software (version 4.11) (Nikon). Fluorescence intensities were quantified from at least 100 cells in ≥ 10 random high-power fields (HPFs) per each sample. Their spectral data were also examined to determine distribution and localization of the protein.

2.12. Cell spindle index measurement

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, cell images were captured using an Eclipse Ti-S phase-contrast inverted microscope equipped with NIS-Elements D software (version 4.11) (Nikon). Using this software, lengths and widths of at least 100 cells in ≥ 10 random HPFs per sample were measured to calculate spindle index [50], [51] as follows.

Spindle index = Cell length / Cell width

2.13. Statistical analysis

The data are reported as mean ± standard deviation (SD) (unless stated otherwise) of measurements obtained from three separate experiments using independent biological samples. Unpaired Student’s t-test was performed to determine difference between two groups, whereas one-way ANOVA with Tukey’s post-hoc test was used for comparing more than two groups of samples. P value < 0.05 was set as the significant threshold.

3. Results

3.1. Effects of trigonelline on the morphology and viability of MDCK cells

The cells were incubated with trigonelline at 1, 10 or 100 µM for 24-h. Cell morphology was examined and dead cells were quantified. The data showed neither obvious morphological changes (Fig. 1A) nor significant increase in cell death (Fig. 1B) induced by any of these concentrations of trigonelline as compared with the untreated control. Therefore, trigonelline at 100 µM was chosen for all following experiments.

Fig. 1.

Optimal concentration for trigonelline treatment. MDCK cells were treated with trigonelline (TRIG) at 1, 10 or 100 µM for 24-h. (A): Cell morphology was examined under an Eclipse Ti-S phase-contrast inverted microscope (Nikon). (B): Percentage of cell death was quantified. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. No significant differences among groups were detected.

3.2. Trigonelline induced changes in levels of several proteins in MDCK cells

From a total of 1006 and 1011 proteins identified from control and trigonelline-treated MDCK cells, respectively, MaxQuant-based quantitative proteomics using nanoLC-ESI-Qq-TOF MS/MS revealed 62 differentially expressed proteins between groups. These include 23 upregulated and 39 downregulated proteins induced by trigonelline (Fig. 2A and Table 1). Discrimination of the sample groups by orthogonal partial least squares discriminant analysis (OPLS-DA) showed the clear separation of the control and trigonelline-treated groups (Fig. 3A). Additionally, permutations could significantly confirm the generated OPLS-DA model (Fig. 3B).

Fig. 2.

Label-free quantitative proteomics and confirmation by Western blotting. After 24-h incubation with/without 100 µM trigonelline (TRIG), cellular proteins were extracted and subjected to in-solution tryptic digestion, nanoLC-ESI-Qq-TOF MS/MS analysis, and label-free quantitative proteomics. (A): A Venn diagram illustrating numbers of total and differentially expressed proteins identified from control and trigonelline-treated cells. (B): Western blotting to confirm the increased level of annexin A2 and decreased level of β-actin induced by trigonelline. GAPDH served as the loading control. (C): Band intensity of each protein was quantified and normalized by that of GAPDH. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. Significant P values are labelled.

Table 1.

Summary of proteins with significantly altered levels in MDCK cells induced by trigonelline.

Protein name	Swiss-Prot ID	Gene symbol	MS/MS identification score	%Cov	No. of distinct/ total matched peptides	MW (kDa)	LFQ intensity (×105a.u.) (Mean ± SEM)		Ratio (Trigonelline/Control)	Pvalue	VIP value
							Control	Trigonelline
10 kDa heat shock protein, mitochondrial	P61604	HSPE1	37.63	73.5	9/9	10.93	57.39 ± 1.42	53.05 ± 0.88	0.9244	0.0192	2.0135
14–3–3 protein eta	P68511	YWHAH	33.75	39.4	6/10	28.21	29.61 ± 0.75	26.77 ± 0.88	0.9042	0.0253	1.9516
26 S proteasome non-ATPase regulatory subunit 1	Q99460	PSMD1	15.16	7.7	5/5	105.84	8.65 ± 0.55	5.66 ± 1.18	0.6547	0.0353	1.7682
40 S ribosomal protein S12	Q76I81	RPS12	151.96	74.2	9/9	14.52	72.85 ± 1.68	66.29 ± 2.43	0.9100	0.0411	1.7912
40 S ribosomal protein S28	Q6QAT1	RPS28	19.55	56.5	5/5	7.84	11.45 ± 1.67	17.83 ± 1.68	1.5573	0.0158	2.0441
40 S ribosomal protein S29	Q6QAP6	RPS29	3.16	26.8	2/2	6.60	2.12 ± 1.40	7.53 ± 1.90	3.5512	0.0356	1.8017
40 S ribosomal protein S4	P79103	RPS4	109.74	45.2	14/14	29.60	57.29 ± 1.19	65.68 ± 2.32	1.1465	0.0054	2.2702
40 S ribosomal protein S8	Q5E958	RPS8	105.71	48.6	9/9	24.21	58.31 ± 1.05	61.39 ± 0.89	1.0529	0.0397	1.8091
40 S ribosomal protein S9	Q6ZWN5	RPS9	30.77	46.9	10/10	22.59	62.65 ± 1.85	68.37 ± 1.20	1.0913	0.0198	1.9884
60 S ribosomal protein L19	Q3T0W9	RPL19	100.21	21.9	6/6	23.47	44.20 ± 0.87	40.66 ± 1.09	0.9200	0.0216	1.9446
60 S ribosomal protein L35	Q3MHM7	RPL35	21.10	29.3	4/4	14.57	12.95 ± 1.47	21.98 ± 2.16	1.6974	0.0032	2.3903
Actin, cytoplasmic 2	P63261	ACTG1	323.31	76.8	1/29	41.79	770.37 ± 11.22	692.01 ± 13.92	0.8983	0.0005	2.6759
Actin-related protein 2	Q5M7U6	ACTR2	23.36	15.0	4/4	44.73	12.98 ± 0.37	11.95 ± 0.29	0.9209	0.0442	1.7199
Actin-related protein 2/3 complex subunit 4	Q148J6	ARPC4	7.78	16.1	3/3	19.67	9.28 ± 0.25	10.32 ± 0.21	1.1112	0.0069	2.2276
ADP/ATP translocase 3	P32007	SLC25A6	22.51	30.9	4/11	32.88	32.35 ± 1.32	36.05 ± 1.07	1.1143	0.0442	1.7542
Alpha-actinin-1	Q7TPR4	ACTN1	16.89	18.6	5/14	103.07	10.34 ± 0.49	12.39 ± 0.57	1.1978	0.0157	2.0352
Annexin A2	Q6TEQ7	ANXA2	323.31	69.3	26/26	38.65	285.46 ± 4.95	321.17 ± 8.84	1.1251	0.0028	2.4381
Annexin A7	P20073	ANXA7	4.06	7.4	3/3	52.74	2.25 ± 0.89	0.00 ± 0.00	0.0000	0.0227	1.9381
Asparagine synthetase [glutamine-hydrolyzing]	Q1LZA3	ASNS	26.46	21.4	1/9	64.22	1.53 ± 0.64	4.35 ± 0.99	2.8391	0.0300	1.8745
Asparagine--tRNA ligase, cytoplasmic	O43776	NARS1	2.88	5.3	2/2	62.94	4.28 ± 0.16	1.34 ± 0.67	0.3126	0.0006	2.6651
ATP synthase subunit alpha, mitochondrial	P19483	ATP5F1A	253.06	48.1	1/23	59.72	109.94 ± 2.12	125.39 ± 2.47	1.1405	0.0002	2.7768
ATP-citrate synthase	Q32PF2	ACLY	36.89	12.5	1/10	119.79	20.64 ± 0.48	19.33 ± 0.32	0.9363	0.0366	1.8374
Basic leucine zipper and W2 domain-containing protein 1	Q9CQC6	BZW1	14.40	17.7	6/6	48.04	0.00 ± 0.00	2.89 ± 1.15	#DIV/0!	0.0229	1.9511
Cell division control protein 42 homolog	Q8CFN2	CDC42	12.96	31.4	4/5	21.26	10.67 ± 0.80	15.83 ± 1.70	1.4845	0.0142	2.0710
Cytosolic non-specific dipeptidase	Q3ZC84	CNDP2	3.58	4.0	1/2	52.66	1.76 ± 0.56	0.00 ± 0.00	0.0000	0.0061	2.2440
DNA replication licensing factor MCM5	P33992	MCM5	6.57	8.9	5/5	82.29	4.77 ± 1.22	1.39 ± 0.92	0.2920	0.0421	1.7688
DNA replication licensing factor MCM6	Q14566	MCM6	7.86	6.0	4/4	92.89	7.63 ± 0.19	3.84 ± 1.22	0.5033	0.0072	2.2345
Eukaryotic translation initiation factor 3 subunit I	Q5E966	EIF3I	7.53	29.8	5/5	36.47	8.94 ± 0.21	4.00 ± 1.58	0.4476	0.0070	2.2145
Far upstream element-binding protein 2	Q92945	KHSRP	22.70	17.0	10/10	73.11	22.09 ± 0.32	20.40 ± 0.43	0.9236	0.0058	2.2686
Glucose-6-phosphate isomerase	P06744	GPI	11.22	5.0	2/2	63.15	2.81 ± 0.90	5.78 ± 0.40	2.0542	0.0081	2.2118
GTP-binding nuclear protein Ran	Q3T054	RAN	53.60	33.3	8/8	24.42	74.35 ± 0.81	68.47 ± 1.15	0.9209	0.0007	2.6611
Heterogeneous nuclear ribonucleoprotein F	Q5E9J1	HNRNPF	44.38	20.5	5/7	45.69	18.54 ± 0.94	15.56 ± 0.69	0.8393	0.0218	1.9753
Heterogeneous nuclear ribonucleoprotein L	Q8R081	HNRNPL	61.14	25.6	11/11	63.96	30.63 ± 0.73	24.65 ± 1.02	0.8049	0.0002	2.7787
Heterogeneous nuclear ribonucleoproteins A2/B1	O88569	HNRNPA2B1	252.84	41.1	13/13	37.40	101.20 ± 2.34	92.46 ± 3.01	0.9136	0.0356	1.7923
Hexokinase-2	P52789	HK2	4.80	3.2	2/2	102.38	4.64 ± 0.59	2.21 ± 0.88	0.4773	0.0353	1.8302
Lamin-B1	P14733	LMNB1	198.28	35.7	4/26	66.79	113.41 ± 2.73	105.36 ± 2.08	0.9290	0.0320	1.8425
Matrin-3	P43243	MATR3	40.77	11.8	1/6	94.62	14.27 ± 0.48	12.76 ± 0.37	0.8942	0.0240	1.9348
Mitochondrial import inner membrane translocase subunit Tim13	Q9Y5L4	TIMM13	4.40	25.3	2/2	10.50	2.27 ± 0.58	0.42 ± 0.42	0.1849	0.0198	1.9990
Myosin-9	Q8VDD5	MYH9	323.31	26.9	38/45	226.37	108.55 ± 2.38	116.03 ± 1.39	1.0689	0.0152	2.0276
NADH dehydrogenase [ubiquinone] 1 alpha subcomplex subunit 13	Q95KV7	NDUFA13	4.11	13.9	2/2	16.67	3.36 ± 0.67	1.12 ± 0.56	0.3339	0.0209	1.9603
Non-POU domain-containing octamer-binding protein	Q15233	NONO	76.26	33.1	11/11	54.23	41.76 ± 0.86	38.28 ± 1.31	0.9167	0.0415	1.7688
Non-specific lipid-transfer protein	P32020	SCP2	172.41	12.4	1/7	59.13	26.36 ± 1.53	33.89 ± 2.34	1.2859	0.0159	2.0250
Peptidyl-prolyl cis-trans isomerase B	P23284	PPIB	100.57	25.9	5/5	23.74	32.18 ± 0.95	35.66 ± 1.03	1.1082	0.0243	1.9118
Peptidyl-tRNA hydrolase 2, mitochondrial	Q3ZBL5	PTRH2	11.39	18.4	3/3	19.29	0.00 ± 0.00	2.07 ± 0.69	#DIV/0!	0.0086	2.1839
Poly(U)-binding-splicing factor PUF60	Q2HJG2	PUF60	22.37	14.7	4/4	57.09	6.28 ± 0.35	4.75 ± 0.63	0.7558	0.0493	1.7012
Probable ATP-dependent RNA helicase DDX17	Q501J6	DDX17	56.20	22.9	9/13	72.40	19.53 ± 0.43	18.00 ± 0.40	0.9212	0.0184	2.0230
Proteasome activator complex subunit 3	P61290	PSME3	15.05	28.3	6/6	29.51	8.71 ± 0.29	7.70 ± 0.26	0.8835	0.0191	1.9863
Proteasome subunit alpha type-2	Q3T0Y5	PSMA2	2.95	14.1	2/2	25.90	0.00 ± 0.00	4.30 ± 1.38	#DIV/0!	0.0067	2.2294
Proteasome subunit beta type-6	P28072	PSMB6	8.34	17.6	2/4	25.36	18.81 ± 0.91	15.81 ± 0.62	0.8404	0.0150	2.0664
Protein arginine N-methyltransferase 1	Q63009	PRMT1	22.63	22.9	6/6	40.52	17.60 ± 0.47	15.83 ± 0.53	0.8994	0.0240	1.9421
Protein PBDC1	Q9BVG4	PBDC1	4.99	9.4	2/2	26.06	9.02 ± 0.42	5.34 ± 1.39	0.5921	0.0222	1.9862
Protein transport protein Sec23B	Q9D662	SEC23B	2.20	3.8	2/2	86.44	0.61 ± 0.40	2.11 ± 0.54	3.4601	0.0408	1.7612
Prothymosin alpha	P06454	PTMA	38.92	34.2	4/4	12.20	14.83 ± 2.11	6.24 ± 3.17	0.4206	0.0383	1.8681
Radixin	P26043	RDX	27.73	20.8	3/11	68.54	4.07 ± 0.53	1.87 ± 0.74	0.4585	0.0283	1.9012
Ras-related protein Rab-1B	Q9H0U4	RAB1B	70.79	54.2	5/11	22.17	28.83 ± 1.14	33.33 ± 1.01	1.1558	0.0095	2.1429
Retinal dehydrogenase 1	P48644	ALDH1A1	16.42	10.6	1/4	54.81	5.28 ± 0.70	2.05 ± 1.05	0.3887	0.0208	1.9586
RuvB-like 1	Q9Y265	RUVBL1	33.88	37.3	13/13	50.23	26.12 ± 0.47	24.56 ± 0.50	0.9402	0.0376	1.8400
Small nuclear ribonucleoprotein F	Q3T0Z8	SNRPF	5.98	24.4	2/2	9.73	5.80 ± 1.12	1.57 ± 1.04	0.2709	0.0138	2.0907
Succinate dehydrogenase [ubiquinone] flavoprotein subunit, mitochondrial	Q8K2B3	SDHA	32.87	8.4	3/4	72.59	12.12 ± 0.22	11.04 ± 0.39	0.9105	0.0277	1.8591
Synaptic vesicle membrane protein VAT-1 homolog	Q99536	VAT1	10.02	13.0	4/4	41.92	3.19 ± 1.01	0.58 ± 0.58	0.1820	0.0402	1.7829
Transcription intermediary factor 1-beta	Q13263	TRIM28	16.61	9.8	6/6	88.55	12.39 ± 0.37	11.14 ± 0.44	0.8992	0.0457	1.7365
Transmembrane emp24 domain-containing protein 9	Q9BVK6	TMED9	55.45	17.4	4/4	27.28	7.20 ± 0.18	8.19 ± 0.22	1.1372	0.0031	2.4035

%Cov = (No. of the identified amino acid residues/No. of amino acid residues in the protein sequence) × 100.

LFQ = label-free quantification.

SEM = standard error of mean.

#DIV/0! = divided by zero.

VIP value = variable influence on projection value (obtained from the OPLS-DA model generated using ropls R package).

Fig. 3.

Discrimination of the sample groups by orthogonal partial least squares discriminant analysis (OPLS-DA). (A): Data distribution and the OPLS-DA scores calculated using the ropls R package to discriminate the control (C, red) and trigonelline-treated (T, blue) groups. The OPLS-DA score plot displays clear separation of the control and trigonelline-treated groups. (B): The 200 random permutations could significantly confirm the generated OPLS-DA model.

From these, upregulation of annexin A2 and downregulation of β-actin were randomly selected to be confirmed by Western blotting. The results confirmed significantly increased level of annexin A2 and significantly decreased level of β-actin in trigonelline-treated cells, consistent with the mass spectrometric data (Fig. 2B and C).

3.3. Biological processes/pathways involved by the trigonelline-induced significantly altered proteins

To predict relevant functions of all these significantly altered proteins induced by trigonelline, their gene symbols were submitted for functional enrichment and reactome pathway analyses using the ClueGO plug-in and Cytoscape software. Functional enrichment analysis showed that the biological processes involved by these trigonelline-induced significantly altered proteins included cellular response to stress, cell cycle and cell polarity (Fig. 4A). Reactome pathway analysis also revealed that cellular response to stress, cell cycle and cell polarity were the three main biological pathways involved by these trigonelline-induced significantly altered proteins (Fig. 4B). Validation of these predictions was then performed using COM crystal as the stress/disease inducer, whereas trigonelline served as the protector/rescuer.

Fig. 4.

Functional enrichment and reactome pathway analyses of significantly altered proteins. All significantly altered proteins induced by trigonelline were subjected to functional enrichment analysis (A) and reactome pathway analysis (B) using the ClueGO plug-in and Cytoscape software. The GO terms of biological processes and individual proteins involved in each biological pathway are labelled.

3.4. Trigonelline prevented COM crystal-induced intracellular ROS overproduction

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, intracellular ROS level was measured using DCFH-DA assay combined with flow cytometry. Fig. 5A shows histogram of cell number (y-axis) and fluorescence intensity (x-axis) after DCFH-DA probing, whereas Fig. 5B demonstrates the quantitative analysis of the data shown in Fig. 5A. The analysis revealed that the intracellular ROS level was significantly elevated by COM crystals compared with the untreated control (Fig. 5). However, pretreatment with trigonelline completely prevented such increase of intracellular ROS level induced by COM crystals (Fig. 5).

Fig. 5.

Trigonelline prevented COM crystal-induced intracellular ROS overproduction. After 24-h pretreatment with/without 100 µM trigonelline (TRIG) followed by 24-h incubation with COM crystals, intracellular ROS level was measured using DCFH-DA assay and flow cytometry. (A): Histogram of the cell number and fluorescence intensity. (B): Percentage of the DCFH-DA-positive cells were quantified from at least 10,000 acquisitions in each sample. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. Significant P values are labelled.

3.5. Trigonelline prevented COM crystal-induced cell cycle shift from G0/G1 to G2/M phase

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, DNA contents were analyzed by propidium iodide staining followed by flow cytometry to determine cell cycle distribution (Fig. 6A). Quantitative analysis of the cells in individual cycles showed that COM crystals caused significant decrease in cell population at G0/G1 phase and increase in cell population at G2/M phase, indicating cell cycle shift from G0/G1 to G2/M phase, as compared with the untreated control (Fig. 6B). However, trigonelline pretreatment successfully prevented such cell cycle shift induced by COM crystals (Fig. 6).

Fig. 6.

Trigonelline prevented COM crystal-induced cell cycle shift from G0/G1 to G2/M phase. After 24-h pretreatment with/without 100 µM trigonelline (TRIG) followed by 24-h incubation with COM crystals, DNA contents were analyzed by propidium iodide staining followed by flow cytometry to determine cell cycle distribution. (A): Histogram of the cell number and fluorescence intensity of each phase of the cell cycle. (B): Distribution of the cell population in G0/G1, S and G2/M phases of the cell cycle were quantified from at least 10,000 acquisitions in each sample. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. Significant P values are labelled.

3.6. Trigonelline prevented COM crystal-induced tight junction disruption

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, transepithelial electrical resistance (TEER) was measured using Millicell-ERS Voltohmmeter, and expression of a tight junction protein (ZO-1) was examined by immunofluorescence staining and Western blotting. The data showed that COM crystals caused significant reduction of TEER (Fig. 7A). Immunofluorescence study revealed that COM crystals significantly suppressed expression level of ZO-1 (Fig. 7B). Moreover, spectral intensity data obviously showed the COM crystal-induced tight junction disruption at the cell borders (Fig. 7C). Western blotting confirmed the reduced level of ZO-1 in the COM-treated cells (Fig. 7D and E). Such markers for tight junction disruption induced by COM crystals were completely preserved at their basal levels by trigonelline pretreatment (Fig. 7).

Fig. 7.

Trigonelline prevented COM crystal-induced tight junction disruption. After 24-h pretreatment with/without 100 µM trigonelline (TRIG) followed by 24-h incubation with COM crystals, TEER and ZO-1 expression were measured. (A): TEER was measured using Millicell-ERS Voltohmmeter. (B): Immunofluorescence staining of ZO-1. (C): Spectral intensity of ZO-1 across the cells (from the area indicated by the red arrow). a.u. = arbitrary unit. (D): Western blotting of ZO-1 and GAPDH (loading control). (E): Band intensity of ZO-1 was quantified and normalized by that of GAPDH. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. Significant P values are labelled.

3.7. Trigonelline prevented COM crystal-induced epithelial-mesenchymal transition (EMT)

After 24-h pretreatment with/without 100 µM trigonelline followed by 24-h incubation with COM crystals, EMT markers (including spindle index, E-cadherin and vimentin) were examined. The analyses revealed that COM crystals caused cell elongation (Fig. 8A), increased spindle index (Fig. 8B) and expression of vimentin (a mesenchymal marker) (Fig. 8C and D), but decreased expression of E-cadherin (an epithelial marker) (Fig. 8C and D), indicating the COM crystal-induced EMT process in renal epithelial cells. All of these EMT markers were successfully preserved by trigonelline pretreatment (Fig. 8).

Fig. 8.

Trigonelline prevented COM crystal-induced EMT. After 24-h pretreatment with/without 100 µM trigonelline (TRIG) followed by 24-h incubation with COM crystals, EMT markers were evaluated. (A): Cell morphology. (B): Spindle index was measured from at least 100 cells in ≥ 10 random HPFs per sample. (C): Western blotting of E-cadherin (epithelial marker) and vimentin (mesenchymal marker), whereas GAPDH served as the loading control. (D): Band intensity of each protein was quantified and normalized by that of GAPDH. Each dot represents each data value, whereas the bar indicates mean ± SD. The data were acquired from three separate experiments using independent biological samples. Significant P values are labelled.

4. Discussion

This study employed label-free quantitative proteomics approach to explore the response of renal cells to trigonelline. The range of trigonelline concentrations tested (1–100 µM) was chosen based on its physiologic range observed in humans after coffee consumption [8], [52], [53]. This range of trigonelline concentrations has also been used in several previous in vitro studies [54], [55], [56], [57]. Cell death analysis, which has been widely used for determination of severe cytotoxic effects from various chemicals, compounds and drugs [58], [59], [60], was applied to address the potential severe cytotoxic effect of trigonelline in the present study. Morphological examination and cell death analysis revealed that trigonelline at 1, 10 and 100 µM had no significant effects on cell morphology and death, consistent with the data reported in other studies indicating that trigonelline at 100 µM or lower concentrations does not cause severe cytotoxicity and significant increase in cell death in mesangial cells, hepatocytes and cardiocytes [61], [62], [63]. Therefore, its concentration at 100 µM was used for all following experiments to gain the maximal effects of this compound while its cytotoxic effects, if any, were minimal.

MaxQuant-based quantitative proteomics using nanoLC-ESI-Qq-TOF MS/MS identified 62 significantly altered proteins induced by trigonelline from over a thousand of proteins identified in renal cells. Quantitative proteomics generally provides a large variable dataset that reflects response of multiple proteins upon stimuli. Multivariate statistical analysis is thus commonly applied to find the meaning of such large variable dataset and to discriminate groups of the samples [64], [65]. The OPLS-DA, one of the multivariate statistical methods, is widely applied in several proteomics [66], [67], [68], [69] and metabolomics [70], [71], [72], [73] studies to discriminate the sample groups. Moreover, the VIP values greater than 1 are widely used to indicate the impact of the quantified proteins in the OPLS-DA model in many proteomics and metabolomics studies [70], [74], [75], [76]. In our study, the VIP values of all significantly altered proteins were > 1 (see Table 1), strengthening their impact on the OPLS-DA model to discriminate the trigonelline-treated group from the control. As mentioned above, the concentration of trigonelline used herein was at physiologic/subtoxic level. Therefore, it was not surprising that the number of significantly altered proteins was relatively small and the degree of changes (in term of fold-change) was relatively low as compared with other studies on chemicals-induced changes in cellular proteome (note that many of these studies do not address the cytotoxic effects from the tested chemicals). We therefore, used the VIP values > 1 (instead of fold-change) as another threshold for determining significant differences in addition to statistical P values. Among the significant differences identified, changes in levels of annexin A2 and β-actin induced by trigonelline were selected to be confirmed by Western blotting based on their highest MS/MS identification score at 323.31 (see Table 1) and the availability of their primary antibodies in our laboratory at the time of investigations. Functional prediction by GO biological process enrichment and reactome pathway analyses consistently guided that these trigonelline-induced altered proteins were involved in cellular response to stress, cell cycle and cell polarity.

According to many lines of evidence, trigonelline has beneficial effects on prevention of several diseases [11], [77]. Our recent study has also reported the direct effects of trigonelline on COM crystals to inhibit kidney stone formation [17]. However, cellular mechanisms underlying its anti-lithiatic effects remain largely unknown. Functional validation of the data predicted from enrichment and reactome analyses, therefore, focused on cellular mechanisms responsible for the anti-lithiatic effects of trigonelline.

Recent in vivo studies have found renal tissue injury, accompanied with ROS overproduction and oxidative stress, in ethylene glycol-induced nephrolithiatic rats [78], [79]. Additionally, recent in vitro studies have confirmed that COM crystals cause renal epithelial cell injury via induction of ROS overproduction and oxidative stress [28], [80], [81]. Our present study has shown the anti-oxidative effects of trigonelline, which completely prevented the increase of intracellular ROS induced by COM crystals. This finding is consistent with a previous study reporting that trigonelline can reduce intracellular ROS production induced by ultraviolet B (UVB) radiation in human foreskin fibroblasts [82]. Similarly, the anti-oxidative effects of trigonelline have also been reported to prevent UVB-induced oxidative damage of DNA contents in mice and human dermal fibroblasts [54]. Moreover, the anti-oxidative effects of trigonelline have been documented in several other in vivo studies using various disease models [13], [83], [84], [85]. These data underscore the anti-oxidative property of trigonelline that may be responsible for its anti-lithiatic effects.

DNA content analysis has revealed that COM crystals induced cell cycle shift from G0/G1 to G2/M phase, indicating cell cycle arrest at G2/M phase. In concordance, several recent studies have reported G2/M phase arrest caused by oxidative stress induced by various stimuli [86], [87], [88]. Additionally, G2/M phase arrest has been found in ischemia-induced proximal renal tubular cell injury [89]. The arrest at G2/M phase is also associated with development of renal fibrosis due to secretion of profibrogenic factors [89]. More importantly, cell cycle shift from G0/G1 to S and G2/M phases have been found in a recent study on scratch- and chemical-induced renal tubular cell injury [90]. These cells with cell cycle shift to S and G2/M phases induce COM crystal adhesion on their surfaces [90], thereby promoting the kidney stone mechanism. Our present study has shown the protective effects of trigonelline on COM-induced cell cycle shift from G0/G1 to G2/M phase, indicating that the anti-lithiatic effects of trigonelline may be mediated, at least in part, via preservation of cell cycle distribution of renal epithelial cells.

Cell polarity is a characteristic of epithelial cells that generally have two poles, apical and basolateral, to regulate and maintain cellular structure and functions. Similar to other epithelial cells, renal epithelial cells exhibit apico-basolateral poles that are separated by tight junctional protein complex [91]. This complex plays crucial roles in establishment and maintenance of renal cell polarity structure and functions. As such, loss or disruption of tight junction contributes to defective cell polarity and altered cellular functions. A recent study has reported that oxalate can induce renal epithelial barrier disruption via inhibition of Wnt/β-catenin signaling [92]. COM crystals also cause tight junction impairment through p38 MAPK signaling and intracellular ROS overproduction [93], [94]. Herein, we confirmed that COM crystals caused tight junction disruption as shown by the declines of TEER and ZO-1 expression. Interestingly, trigonelline could completely preserved tight junction structure and function. These data were consistent with other evidence reporting that bioactive compounds with anti-oxidant properties can alleviate stimuli-induced damages of tight junction complex [95], [96], [97].

Kidney stone disease also increases the risk for development of CKD [98]. Renal fibrosis is an important feature of CKD. One of the cellular mechanisms triggering renal fibrosis is EMT [99]. This transitional cellular process is characterized by loss of epithelial properties and gain of mesenchymal features of the cells [99]. A recent study has reported the EMT induction together with ROS accumulation in a mouse model of pulmonary fibrosis caused by high-glucose condition [100]. Herein, we found that COM induced EMT in renal epithelial cells as evidenced by the increased spindle index, decreased epithelial marker (E-cadherin) and increased mesenchymal marker (vimentin). Again, trigonelline successfully prevented these EMT features triggered by COM crystals. Our findings were consistent with the data reported in previous studies indicating that other natural bioactive compounds (i.e., epigallocatechin-3-gallate, caffeine, rhein and curcumin) with anti-oxidative and anti-inflammatory properties also exhibit the anti-fibrotic effects [50], [101], [102]. Additionally, a recent study on mesangial cells has revealed that trigonelline can reverse high-glucose-induced fibrosis by inhibition of Wnt signaling [55]. A previous study in neonatal diabetic rats has found that trigonelline can reduce renal fibrosis by suppressing oxidative stress [103]. Nevertheless, the evidence showing the advantages of trigonelline to prevent EMT process is limited. Only one recent study has shown that trigonelline can prevent high-oxalate-induced EMT progression [104]. These data implicate that, in addition to the protective mechanisms against kidney stone formation, trigonelline also prevents nephrolithiasis-associated CKD by preventing the EMT fibrotic mechanisms.

Finally, limitations of our present study should be also mentioned. Although a set of significantly altered proteins could lead to functional prediction followed by experimental validation, the number of significantly altered proteins was relatively small as compared with those reported in other studies. This was mainly due to the physiologic/subtoxic concentration of trigonelline selected and the mass spectrometric modality employed. Using the greater concentration of trigonelline and more sensitive mass spectrometric modality would enhance the identification of the significantly altered proteins in renal cells induced by trigonelline. However, the potential cytotoxicity of the greater concentration of trigonelline must be considered as it may interfere with the data interpretation. Additionally, the study was done with only a single time-point (24 h) for trigonelline treatment. Serial changes of the cellular proteome at various time-points would yield more in-depth insights and dynamicity of the trigonelline-induced changes in cellular proteome and function. Furthermore, we examined changes only in MDCK renal cells, whereas analyses on other renal cells (i.e., from other species and/or from different segments of nephron) would yield much greater dimension of the data. Finally, there are several other bioinformatic tools that can be used to enhance the interpretation and/or prediction of the proteome data. Having done so, the understanding of the biological effects of trigonelline in renal cells would be much clearer.

In summary, our present study has reported that trigonelline causes significant changes in levels of 62 (23 upregulated and 39 downregulated) proteins in MDCK cells. Functional enrichment and reactome pathway analyses reveal that these trigonelline-induced significantly altered proteins are related to stress response, cell cycle and cell polarity. Functional investigations confirm that trigonelline prevents COM crystal-induced renal cell deteriorations by inhibiting crystal-induced overproduction of intracellular ROS, cell cycle shift from G0/G1 to G2/M phase, tight junction disruption, and EMT in renal epithelial cells. These findings provide cellular mechanisms and convincing evidence for the renoprotective effects of trigonelline, particularly in kidney stone prevention.

CRediT authorship contribution statement

PPe, WB, PPu, SS, and VT designed research; PPe, WB, PPu, and SS performed experiments; PPe, WB, PPu, SS, and VT analyzed data; PPe and VT wrote the manuscript; All authors reviewed and approved the manuscript.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material

.