Identification of GPR35-associated metabolic characteristics through LC-MS/MS-based metabolomics and lipidomics

G protein-coupled receptor 35 (GPR35) has gained increasing attention as a promising target in treating inflammatory and gastrointestinal tract conditions, cardiovascular diseases, and cancer. Metabolites including kynurenic acid, lysophosphatidic acids, chemokine 17, and 5-hydroxyindole acetic acid have been suggested to be endogenous ligands of GPR35. However, little is known regarding the downstream metabolic characteristics upon GPR35 regulation. Herein, four GPR35 interventions in cell models, comprising GPR35 knock-down, over-expression, activation


In brief
G protein-coupled receptor 35 (GPR35) interventions had a significant impact on the abundance of downstream metabolites and lipids, particularly affecting those involved in β-oxidation and phosphatidylethanolamine pathways.

Highlights
• Four G protein-coupled receptor 35 (GPR35) intervention cell models exhibit significant alterations in the levels of 75 metabolites and 204 lipids.

INTRODUCTION
More than one-third of drugs are believed to target G protein-coupled receptors (GPCRs) [1].However, the ligands and functions of many GPCRs are not well understood; G protein-coupled receptor 35 (GPR35) is one such "orphan" receptor [2].The GPR35 gene is located on chromosome 2q37.3 in human [2] and produces two variants (GPR35a and b) [3].This gene is highly expressed in the gastrointestinal tract and immune system, and is moderately expressed in the whole brain, skeletal muscle, liver, and heart [4].GPR35 has gained increasing attention as a crucial target, owing to its broad pathophysiological association with many diseases.GPR35 plays a substantial role in intestinal homeostasis [4].GPR35 −/− mice treated with dextran sulfate sodium show more severe colitis syndrome than wild-type mice [5].Activation of the GPR35 pathway promotes colorectal tumor growth [6,7].Furthermore, GPR35 has been reported to act in various subpopulations of immune cells; e.g., activated GPR35 decreases interleukin-4 secretion in invariant natural killer T cells [8], promotes neutrophil recruitment [9], and induces cell-cycle arrest in monocytes [10].GPR35 is also associated with cardiovascular diseases.Genomewide association studies have indicated that GPR35 is significantly up-regulated in patients with heart failure [11].Activation of GPR35 protects against ischemia during myocardial infarction [12], and GPR35 −/− relieves angiotensin II-induced hypertension [13].In addition, GPR35 has been implicated in pain perception [14], cancer [15], and type 2 diabetes [16].GPR35 activates G s , G i , G q , and G 12/13 , and regulates other signaling pathways through second messengers (e.g., cyclic adenosine monophosphate, Ca 2+ , and diacylglycerol), thereby exerting physiological functions [4].GPR35 alleviates ulcerative colitis (UC) through complementary mechanisms; i.e., GPR35 inhibits NOD-like receptor thermal protein domain associated protein 3 over-activation and thus relieves inflammation by reducing the level of interleukin-1β [17], GPR35 also increases the expression of fibronectin and integrin-α5β1, thus promoting the repair of colon mucosa [18].Furthermore, GPR35 promotes lipid metabolism through various mechanisms.Agudelo et al. [19] have reported that GPR35 increases levels of peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α) and regulator of G-protein signaling 14, thereby stimulating lipid metabolism and anti-inflammatory gene expression in adipose tissue.In contrast, Nam et al. reported that GPR35 inhibits liver X receptor (LXR), thereby reducing lipid accumulation through the p38 mitogen-activated protein kinase and c-Jun N-terminal kinase pathways in hepatocytes [20].In addition, GPR35 activates Ras homolog gene family member A/Rho-associated coiledcoil kinase signaling and consequently plays a role in multiple cardiovascular diseases [21,22]; interacts with Na + /K + -ATPase and consequently promotes colorectal cancer [6,7]; and recruits β-arrestin-2 [23,24], which acts as a scaffold in signal transduction and is associated with receptor desensitization, and the c-Jun, protein kinase B and extracellular-regulated kinase 1/2 (ERK1/2) pathways [25].
GPR35 is strongly associated with metabolic regulation in various physiological and pathological conditions.GPR35 is a member of the metabolite-sensing GPCRs, which bind various metabolites and transmit signals important for proper immune and metabolic functions [26].Furthermore, GPR35 may act as an integration node in conditions in which inflammation skews metabolism, given that it is expressed in adipose tissue, immune cells, and the gastrointestinal tract, and that metabolic disorders often coincide with states of chronic inflammation [27].Moreover, activation of GPR35 promotes lipid metabolism in skeletal muscle [28], adipose tissue [19], and liver cells [20,29].However, little is known regarding the downstream metabolic characteristics upon GPR35 regulation.Identification of GPR35-associated metabolic characteristics would increase understanding of the potential mechanisms and functions of GPR35 in various physiological and pathological conditions.

Cell culture
The HCT116 cell line was obtained from Cobioer Biosciences (Nanjing, China), and was cultured in DMEM supplemented with 10% fetal bovine serum and 0.1% penicillin-streptomycin.Cells were cultured at 37°C in a humidified atmosphere containing 5% CO 2 .The HCT116 cell line was authenticated at Zhong Qiao Xin Zhou Biotechnology (Shanghai, China), with sample code 20170713-06.

Cell transfection
Lentivirus for gene silencing and over-expression of human GPR35 were purchased from GenePharma (Shanghai, China).The lentiviral vector system contained green fluorescent protein and puromycin resistance genes.Cells at 40%-60% confluence were transduced with lentivirus at a multiplicity of infection of 10, and incubated with medium containing 5 μg/mL polybrene (GenePharma, Shanghai).After 24 h, the medium was replaced with fresh culture medium.After 72 h, cells positive for green fluorescent protein were observed under a fluorescence microscope (Nikon, Tokyo, Japan), then selected with medium containing 2 μg/mL puromycin for another 48 h.The oligonucleotide sequences of the lentiviral vectors are listed in Table S1.

Analysis of GPR35 internalization
The quantification of receptor internalization was performed by measurement of specific-antibody-tagged receptors on cell surfaces through flow cytometry.Cells were exposed to drugs or vehicles for 48 h, then collected by centrifugation (1,000 rpm, 5 min) (Eppendorf, Hamburg, Germany).Subsequently, cells in the treatment groups were incubated with anti-GPR35 (1:300, Proteintech, Chicago, IL, USA), whereas cells in the negative control group were incubated with 3% bovine serum albumin in phosphate-buffered saline (PBS) buffer at 4°C for 2 h.All groups were subsequently incubated with secondary antibody Alexa Fluor ® 488 conjugate anti-rabbit IgG (H+L) (1:300, Cell Signaling Technology, Boston, MA, USA) at 4°C for 1 h.The cells were then fixed with 4% formaldehyde.The fluorescence intensity of 1 × 10 4 cells per sample was measured with a NovoCyte system (Agilent, California, United States).The percentage of internalized receptors was calculated from surface receptor fluorescence values (F) according to the following equation:

mRNA preparation and reverse transcription polymerase chain reaction
GPR35 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression was analyzed with a standard reverse transcription polymerase chain reaction (RT-PCR) protocol.Total RNA was extracted from cells with an RNAiso Plus Kit (Takara Biotechnology, Dalian, China).RNA concentrations were detected with a Nano-Drop 2000 instrument (Thermo Fisher Scientific, Waltham, MA, USA).Subsequently, RNA (1 μg) was reverse transcribed to complementary DNA with a PrimeScript™ RT reagent kit (Takara Biotechnology).RT-PCR was performed with SYBR Green I Master Mix (Roche, Basel, Switzerland) on a LightCycler 480 instrument (Roche).
The relative expression levels of GPR35 were calculated with the 2 −ΔΔCT method, on the basis of normalization to GAPDH.The primer sequences for RT-PCR are listed in Table S2.

Western blotting analysis
The protein expression of GPR35, ERK1/2, phospho-ERK1/2 (pERK1/2), and α-tubulin was analyzed with a standard western blotting protocol.Cells were lysed with radioimmunoprecipitation assay buffer (Beyotime Biotechnology, Shanghai, China) containing 0.1% phenylmethylsulfonyl fluoride (Thermo Fisher Scientific) on ice, and total protein was extracted.Protein concentrations were measured with a bicinchoninic acid protein assay kit (Beyotime Biotechnology).The lysates were mixed with sample loading buffer (Beyotime Biotechnology) and denatured at 95°C.Protein samples (30 μg) were separated by sodium dodecyl sulfatepolyacrylamide gel electrophoresis with 8% acrylamide gels and transferred to polyvinylidene difluoride membranes (Massachusetts, United States) (0.2 μm) with an electroblotting apparatus (Tanon, Shanghai, China).The membranes were blocked with 5% (w/v) nonfat milk for 2 h at room temperature and incubated with primary antibodies at 4°C overnight.After the membranes were washed three times with PBS containing 0.1% Tween 20, the immunoreactive bands were incubated with secondary antibodies conjugated to horseradish peroxidase for 2 h at room temperature.The immunoreactive bands were subsequently visualized with enhanced chemiluminescence (Massachusetts) with a Tanon 5200 chemiluminescence imaging system (Tanon, Shanghai, China).Relative protein expression was calculated through densitometric analysis in ImageJ software.The antibodies used for western blotting analysis are listed in Table S3.S4).The chemical derivatization procedure (Dns-Cl and Dns-PP labeling) was as reported in our previous study [31] and is described in the Extended Methods section in the Supporting Information.A Shimadzu Nexera UPLC system interfaced with an 8060 triple quadruple mass spectrometer (Shimadzu, Kyoto, Japan) was used for metabolomics analysis.Details regarding instrument conditions, columns, mobile phases, and MS parameters can be found in the Extended Methods section of the Supplementary Information.Another aliquot of cells was prepared for total protein assays (Beyotime Biotechnology), and the results were used for metabolite normalization.

Sample preparation.
A total of ∼10 7 cells were extracted by liquid-liquid extraction with a MTBE/ MeOH/H 2 O system.Specifically, 500 μL of cold MeOH was added to the cell culture dish, the contents of the dish were transferred into a 4 mL tube, 1 mL of MTBE was added, and the mixture was vortex mixed for 30 seconds.Subsequently, the mixture was vibrated for 30 min, and 250 μL of H 2 O was added and vortex mixed for 30 s to form a two-phase system.After equilibration for 10 min on ice, the supernatant was collected after centrifugation (14,000 rpm, 10 min, 4°C).QC samples were prepared by pooling equal aliquots of each sample.All samples were dried and stored at −80°C before liquidchromatography tandem mass spectrometry (LC-MS/MS) analysis.Similarly, an aliquot of cells was prepared for total protein assays (Beyotime Biotechnology), and the results were used for lipid normalization.2) variables with a relative standard deviation > 15% in QC samples were removed; and 3) missing values were imputed with half the minimum value presented in the dataset.

Statistical analysis
Statistical analysis was performed in GraphPad Prism 8.0 software (GraphPad Software Inc., La Jolla, CA, USA).All experiments were repeated at least three times independently, and the results are presented as mean ± standard deviation unless otherwise specified.Independent unpaired two-tailed Student's t-test was used to evaluate differences between two groups, and multiple group comparisons were analyzed with oneway analysis of variance with Bonferroni correction.P < 0.05 was considered statistically significant.

Establishment of GPR35 intervention cell models
GPR35 is highly expressed in the gastrointestinal tract and is more highly expressed in colorectal cancer cells [33].In this study, the HCT116 cell line was used to establish several GPR35 intervention models, because of the high GPR35 protein expression.First, we used lentiviral transduction to regulate GPR35 expression; consequently, both the mRNA (Figure 1A) and protein (Figure 1B) expression levels of GPR35 were significantly altered.Second, we applied the relatively exclusive ligands pamoic acid (PA) or ML194 to activate or inhibit GPR35, respectively.As shown in Figure 1C, the receptor internalization rate increased with PA treatment in a dose-dependent manner, whereas ML194 treatment at 10 μM strongly antagonized this effect.In addition, the phosphorylation of ERK1/2, which is known to be downstream of GPR35, was also increased by PA treatment or decreased by ML194 treatment (Figure 1D), although GPR35 mRNA and protein expression were unchanged (Figure 1D, E).Thus, the GPR35 intervention models were established successfully, and were considered suitable for further metabolomics and lipidomics analysis.

Screening of GPR35-associated metabolites through targeted metabolomics analysis.
Targeted metabolomics analysis of GPR35 intervention cells was performed to identify metabolites differentially present between the control group and the model groups including GPR35 knock-down (KDNC vs KD), over-expression (OENC vs OE), activation (C vs PA) and inhibition (C vs ML).As shown in the volcano plots (Figure S1), the GPR35 interventions induced significant changes in the metabolic profile.A total of 75 metabolites meeting the standards of fold change > 1.5 and adjusted P < 0.05 were selected.As shown in the heat maps (Figure 2A-D), the metabolic characteristics significantly differed between the model groups and controls.

HAA and Ura are sensitive to all GPR35 interventions.
To obtain the metabolites overlapping among interventions, we imported all GPR35-associated metabolites in the four groups into VENNY 2.1 and generated a Venn diagram (Figure 3A).Levels of two metabolites, 3-hydroxyanthranilic acid (HAA) and uracil (Ura), significantly differed across all conditions, thus suggesting the sensitivity of these metabolites to GPR35 interventions.However, HAA and Ura decreased when GPR35 was activated or inhibited, and increased when GPR35 was knocked down or over-expressed (Figure 3B).These findings were inconsistent with the traditional concept in which opposite interventions result in opposite changes.We searched the literature and found a similar phenomenon reported in a study of the relationship between GPR35 and hypoxia-inducible factor-1α (HIF-1α) [34] or Na + /K + -ATPase [6,7].Specifically, HIF-1α expression is up-regulated when GPR35 is activated or inhibited, and intracellular Ca 2+ concentrations increase when GPR35 is knocked out or activated.

Metabolite variation trends.
We further analyzed the variation trends of metabolites under the different interventions and graded the metabolite levels (Table S6), on the basis of two main premises 1) GPR35 knockdown and over-expression causing opposite trends, and 2) GPR35 activation and inhibition causing opposite trends.Level 1 metabolites meeting the above two premises showed the same trend in GPR35 knock-down and inhibition.Level 2 metabolites meeting the above two premises showed the same trend in GPR35 knockdown and activation.Level 3 metabolites met only one of the two premises.Level 4 metabolites met neither premise.Level 1-3 metabolites and their variation trends are shown in Figure 4. Kynurenic acid (KA) was the only level 1 metabolite: this endogenous agonist of GPR35 increased when GPR35 was inhibited or knocked down, and decreased when GPR35 was activated or over-expressed (Figure 4A).These findings suggested that GPR35 regulates KA through negative feedback, given that KA is usually considered an endogenous ligand of GPR35.Adenosine (Ado), a level 2 metabolite, decreased when GPR35 was activated or knocked down (Figure 4B).In addition, L-isoleucine, octanoic acid, and L-aspartic acid were identified as level 3 metabolites (Figure 4C, E).

GPR35 interventions significantly perturb fatty
acid β-oxidation.GPR35-associated metabolites were analyzed by compound class (e.g., fatty acids, acylcarnitines, and amino acids).As shown in Figure 5A, fatty acids and acylcarnitines were dominant among all metabolite classes.Enrichment analysis based on GPR35-associated fatty acids and acylcarnitines indicated that fatty acid β-oxidation (FAO) of very long chain fatty acids showed the greatest perturbation (Figure 5B).analysis was performed to identify lipids differentially present across four cell models.On the basis of previous literature [32], 350 lipids were monitored in this study (Table S5).QC samples were included in the analytical batch for data quality evaluation.All lipids in the QC samples showed little variation, with a relative SD <15% (Figure S2A), and principal component analysis demonstrated that the QC samples were closely clustered (Figure S2B, C), thus suggesting the good stability and reproducibility of the method.GPR35-associated lipids were screened according to the same standards used in the metabolomics, and volcano plots were obtained (Figure S3).Consequently, 204 differentially present lipids were characterized as GPR35-associated lipids, Similarly to the findings regarding the metabolites, lipid metabolism was significantly altered in the model groups compared with the controls.In particular, most lipids decreased with GPR35 activation (Figure 6D).

No lipids showed significant changes in abundance under all interventions.
Overlapping lipids under the different interventions were analyzed similarly to the metabolomics analyses.As shown in the Venn diagram (Figure 7), no lipids showed significant changes in abundance under all interventions.In contrast, 125 lipids showed significant changes in abundance when GPR35 was activated, in agreement with the heat map analysis, thereby suggesting that GPR35 activation caused more significant changes in lipid metabolism than GPR35 inhibition.Furthermore, 35 lipids showed changes in abundance when GPR35 was knocked down or activated; therefore, opposite interventions might not necessarily cause opposite change trends in lipid levels.

Lipid variation trends.
Similarly, we analyzed the variation trends in lipid levels by applying the same criteria described above in the metabolomics section.Level 1-4 lipids and their relative ratios are shown in Table S7.Among them, 56 lipids were identified as level 1, including primarily 20 phosphatidyl ethanolamines (PEs), 13 phosphatidylcholines, and 9 sphingomyelins.As with KA, these lipids showed increased concentrations when GPR35 was inhibited or knocked down, and decreased concentrations when GPR35 was activated or over-expressed.Eight lipids graded as level 2 showed reverse changes with intracellular Ca 2+ .Another 56 lipids were classified as level 3, most of which were phosphatidylcholines.The concentrations of these lipids significantly decreased when GPR35 was activated and increased when GPR35 was inhibited, whereas no significant change was observed when the expression of GPR35 was regulated.

PE metabolism is significantly perturbed by GPR35 interventions.
We analyzed the changes according to lipid subclass by referring to the existing literature.Triglyceride (TGs) significantly decreased with GPR35 knock-down and increased with GPR35 over-expression (Table S8).However, this finding must be validated by inclusion of more TGs in the analysis in the future.Moreover, PEs significantly increased with GPR35 inhibition or knock-down, and significantly decreased with GPR35 activation or over-expression (Figure 8, Table S8).Thus, PE metabolism was significantly perturbed by the GPR35 interventions.

DISCUSSION
GPR35 is closely associated with various pathophysiological conditions and has been demonstrated to be a crucial target for therapeutic interventions.However, little is known regarding the metabolic characteristics of GPR35.In some diseases, the protein levels of GPR35 have been reported to be altered; for example, GPR35 protein is up-regulated under inflammatory challenge [35] or hypoxia (such as myocardial infarction) [21], and is down-regulated in patients with UC [36].In some cases, however, only GPR35 activity is altered; for instance, GPR35 is involved in mediating ischemic protection [12] and neutrophil recruitment [9] activated by KA or 5-Hydroxyindoleacetic acid.Therefore, we constructed four cell models with regulation of GPR35 expression or activity, to understand the related metabolic changes.
Our findings indicated that FAO was significantly perturbed by the GPR35 interventions.To date, no direct link between GPR35 and key enzymes in FAO has been reported.Ni et al. [37] have found that activated HIF-1α causes down-regulation of CPT1 in HCT116 cells, thus resulting in the accumulation of fatty acids and lipids.Moreover, HIF-1α is activated by GPR35 activation or inhibition [34].Therefore, we hypothesized that GPR35 might affect FAO by activating HIF-1α.In contrast, GPR35 might modulate the concentrations of relevant metabolites by affecting classic FAO targets, such as peroxisome proliferator-activated receptor [38], PGC-1α [19,39], and LXR [29,39].For example, activated GPR35 up-regulates PGC-1α [19], and PGC-1α in turn increases FAO to meet energy requirements [39].
Notably, we observed that PE metabolism was perturbed by the GPR35 interventions.PEs are important components of biological membranes, and have diverse roles in cellular functions such as autophagy [40], ferroptosis [41], and immune cell differentiation [42].Dysregulation of PEs has been implicated in many metabolic diseases, such as atherosclerosis, insulin resistance, and obesity [43,44], as well as several chronic diseases, such as Alzheimer's disease, Parkinson's disease, and nonalcoholic liver disease [44,45].GPR35 has also been implicated in type 2 diabetes mellitus [26,28] and nonalcoholic liver disease [29], among others.Furthermore, GPR35, through up-regulation of PGC-1α, significantly affects the expression of genes associated with lipid metabolism, thermogenesis, and anti-inflammation in adipose tissue [19].Moreover, activation of GPR35 inhibits lipid accumulation through LXR inhibition in hepatocyte [20,29].We speculate that these functions of GPR35 might be involved in the regulation of lipid metabolism, particularly that of PE.
Beyond FAO and PE metabolism, we identified specific metabolites sensitive to GPR35 interventions.HAA and Ura decreased with interventions altering GPR35 activity and increased with interventions altering GPR35 expression.These findings were consistent with the changes in HIF-1α [34] or Na + /K + -ATPase reported in prior studies [6,7].Other metabolites, such as KA and Ado, exhibited opposite changes following GPR35 knock-down and over-expression, as well as opposite changes following GPR35 activation and inhibition.The KA-GPR35 axis is involved in various physiological and pathological processes, such as inflammation, analgesia, and cardiovascular system diseases [4,12,17].Ado, similarly to GPR35, is significantly up-regulated under stress states, such as inflammation, ischemia, and hypoxia [46].
This study has several limitations, and our findings require further exploration.For example, experiments are necessary to confirm the links between GPR35 and FAO or PE metabolism.Future investigations may include 1) measurement of all metabolites involved in FAO or PE metabolism in the four cell models to confirm the relationship between GPR35 intervention and these two pathways; 2) exploration of the changes in expression or activity of key enzymes/genes in FAO or PE metabolism under GPR35 interventions; and 3) further validation of through the regulation of related enzymes/genes under GPR35 interventions.In addition, our experiments were performed solely in a cell model, which cannot fully represent the metabolic regulation of GPR35 in vivo; consequently, animal experiments are necessary to confirm our findings.

CONCLUSIONS
GPR35 has widespread pathophysiological associations with many diseases and has gained increasing attention as a promising treatment target.However, the study of GPR35 remains in its infancy.To our knowledge, this study is the first to report the metabolic characteristics of GPR35, according to targeted metabolomics and  pseudotargeted lipidomics.Many metabolites and lipids showed marked changes in abundance under the different GPR35 interventions, among which FAO and PE metabolism were significantly perturbed.Although further experiments are required to validate our findings, our data may promote understanding of the metabolic characteristics of GPR35.